Two years ago, iffy results from a massive Phase 3 trial led a Canadian drug company to scrap development of its anti-amyloid therapy for Alzheimer disease. Today, seniors can purchase the very same compound, rebranded as a “science-based natural health product,” right off the shelves of Canadian drug stores or from their home computers via the Internet. The field at large could now be witnessing variations of this theme, played out more subtly through the appearance of AD’s first “medical foods.” Earlier this spring, Ketasyn (aka AC-1202)—a ketogenic drink powder that targets metabolic deficiencies associated with AD—hit the market under the trade name AxonaTM based on Phase 2 data reported initially in 2007 (see ARF related conference story) and published last month (Henderson et al., 2009). Meanwhile, recruitment proceeds apace for U.S. and European Phase 3 trials of the multi-nutrient drink SouvenaidTM (aka FortasynTM Connect in powder form), which purportedly boosts the brain’s supply of synaptic membrane constituents (see ARF related conference story).

“This is going to be in the limelight,” said Maria Carrillo of the Alzheimer’s Association. “People are going to be looking at dietary supplements and medical foods more and more, especially with the potential epidemic we're facing with AD.” An estimated 35 million people are living with dementia worldwide, and if present trends continue, the number with AD is projected to nearly double every two decades, reaching 65 million by 2030, according to a new report released last month (see ARF related news story).

Defined by the U.S. Food and Drug Administration in 1988, medical foods are a special class of therapeutic agent that fall somewhere in between unregulated food supplements and FDA-approved drugs (see Morgan and Baggott, 2006). Unlike dietary supplements, which are sold over the counter and intended for healthy people, medical foods target people with a specific disease and generally require a doctor’s prescription. However, a key difference from drugs is that medical foods do not undergo pre-market FDA review or approval, and do not have to be registered with the agency (see FDA Guidance for Industry).

With no FDA efficacy claim, how do consumers know that a medical food will do them any good? “The short answer is that consumers don’t know,” FDA supervisory nutritionist Sue Anderson wrote in an e-mail to ARF (see Q&A below). “Manufacturers should have information to substantiate that claims on their labels are truthful and not misleading. However, the claims are not reviewed by the FDA.”

Regulation Loophole
Therein lies the catch—or lure—depending on one’s perspective. Because medical foods consist of ingredients “generally recognized as safe” (GRAS), they escape much of the safety monitoring normally required for drugs. This, combined with the lack of clear standards for demonstrating efficacy, make the “medical food” designation appealing to investors looking to send their companies’ products on a quicker, cheaper route to market.

Take, for example, AC-1202, a form of medium-chain triglycerides being developed for AD by Accera, a biotech company in Broomfield, Colorado. The compound gets converted by the liver into ketone bodies, which enter the brain to provide an alternative energy source for neurons that no longer use glucose properly. “For the investors, one of the big appeals of our approach to AD is that we’re basically using a food ingredient to generate ketosis,” Sam Henderson, Accera’s executive director of research, told ARF. “You don’t have to go through the normal long drug development process. You can very quickly come to market with something like a medical food.”

In a small proof-of-concept study involving 20 people with mild to moderate AD, those who drank a milkshake containing emulsified AC-1202 had higher blood levels of ketone bodies compared with participants on an isocaloric placebo drink. The AC-1202 patients also had better ADAS-Cog scores several hours after drinking the milkshake, but only if they lacked the ApoE4 allele (Reger et al., 2004). That trial, published with Suzanne Craft at the University of Washington, Seattle, was done on angel financing, Henderson said. The company did secure some funding from venture capitalists, but “they didn’t have the monies to do the long-term development process,” Henderson said. “Their real motivation was rapid development of the product since our main ingredient is already recognized as a safe food ingredient.”

Though similar on their safety profiles, medical foods have one up on food supplements in the area of labeling. Whereas supplements are restricted to structure/function claims (e.g., “protects memory function”), medical food labels can mention specific diseases. According to the company’s website, AxonaTM (trade name for the AC-1202-containing drink) “is a specially formulated medical food intended for the clinical dietary management of the metabolic processes associated with mild to moderate AD.”

What’s in a Label?
This description raises eyebrows—mostly for what it does not claim. While the label indicates that the product alters metabolic problems associated with AD, “it doesn’t actually say whether it benefits AD patients or improves memory or anything like that,” said Douglas Galasko of the University of California, San Diego, in an interview with ARF. “In addition to changing some metabolic problem, there has to be a clinically meaningful outcome.”

Thus far, though, the field may lack scientific evidence to make this a reasonable expectation. It’s the classic question of association versus causation, Galasko said, noting weight loss as an example. “Most people with AD lose weight,” he said. “But they’re not developing dementia because they lose weight.” Likewise, Galasko questions whether addressing a metabolic deficiency associated with AD would actually slow the course of disease. Positron emission tomography (PET) studies have long established glucose hypometabolism in AD-affected brain areas. However, “Is that because the brain has a glucose problem or because synapses aren’t working properly and in fact there’s an appropriate amount of glucose metabolism per functional synapse?” he asked.

Craft, who ran Accera’s initial pilot study suggesting that Axona improved cognition in E4 non-carriers, does not find this chicken-or-egg issue so critical. “Even if energy issues were secondary, that doesn’t mean that supplying energy couldn’t be helpful,” she told ARF. She found the recent data promising (Henderson et al., 2009). “The fact that it was consistent with the pilot study (Reger et al., 2004) suggests, to me, that this is a real effect. It looked like there was some measurable benefit for patients who didn’t have the E4 allele. Those folks showed a three- to four-point ADAS-Cog benefit relative to placebo,” Craft said.

Nevertheless, the study failed to meet its primary endpoint—improved ADAS-Cog scores at day 90, relative to placebo, in treated participants at large. There was a meaningful ADAS-Cog signal compared to placebo at 45 days, but this effect largely dissipated by 90 days. “That’s worrisome,” Galasko said. “The FDA would not license a drug based on 45-day effects that were no longer present at 90 days.” In a phone interview with ARF, Henderson said the company had never filed an IND (Investigative New Drug application) and was “always planning on something like a medical food.” However, Steve Orndorff, Accera’s founder and CEO, spoke of different intentions for Axona in another story.

Silver Lining in ApoE4 Effects?
Craft agrees that the company still “has not done the definitive Phase 3 study” and thinks Axona’s apparent E4 differential benefit deserves further research. “It’s a fairly low-risk product. There’s GI upset (i.e., diarrhea in 24.4 percent of the treated participants versus in 13.6 percent of the placebo group), and of course that’s always a concern when used in a prolonged fashion in an older population. But other than that, there aren’t any identified serious adverse events associated with it. It seems they did show fairly robust benefit for a subgroup of folks,” Craft said. “I’d just like to see them test it one more time in a study that was designed and powered to specifically test this E4 differential response.” PET studies led by Eric Reiman, who heads the Arizona Alzheimer’s Consortium and directs the state’s Banner Alzheimer’s Institute in Phoenix, have shown that ApoE4 gene dose correlates with lower glucose metabolism (Reiman et al., 2005) and more amyloid-β (Reiman et al., 2009 and ARF related news story) in AD-affected brain areas.

Possible E4 effects were also buried within pre-planned exploratory analysis of data from the Alzheimer’s Disease Cooperative Study’s (ADCS) trial of docosahexaenoic acid (DHA). In this study, 402 mild to moderate AD patients at 51 U.S. sites took two grams of DHA daily for 18 months. This treatment had no effect on participants’ scores, relative to placebo, on the ADAS-Cog or the Clinical Dementia Scale Sum of Boxes (CDR-SOB), principal investigator Joseph Quinn, Oregon Health and Sciences University, Portland, reported in July at the International Conference on Alzheimer’s Disease (ICAD) in Vienna. However, when the data were re-analyzed in participant subsets defined by ApoE genotype, the researchers found that E4 non-carriers who received DHA declined more slowly on the ADAS-Cog and showed a similar trend on the Mini-Mental State Examination (MMSE). DHA treatment did not help people who had the E4 gene.

DHA supplements did seem to bring modest cognitive benefit in a six-month Memory Improvement with DHA Study (MIDAS) by Martek Biosciences Corporation, Columbia, Maryland, the primary supplier of algal DHA for supplementation. However, unlike the ADCS trial, which involved early AD patients, the MIDAS study enrolled healthy seniors with mild memory complaints. Relative to the placebo group, those who took 900 mg daily DHA capsules made fewer errors on the CANTAB Paired Associate Learning (PAL), an episodic memory test, compared to when they began the study. Some may interpret these findings to suggest that DHA could be more effective as a preventive therapy rather than a treatment for AD. “I wouldn’t jump to that one just yet,” Quinn told ARF. “The effects in the MIDAS study were pretty modest. It’s hard to know the clinical significance of those modest effects. The participants were not selected to represent a prodromal population.”

Discussion is underway for a future ADCS trial in people with mild cognitive impairment (MCI). “We want to look at a group of people who are at really high risk for AD, and look for evidence that DHA intervention could have an effect on outcomes that are more clinically meaningful,” Quinn said. Besides ADAS-Cog, these could include measures that more reliably capture the transition to dementia. The trial would also include stratification for ApoE genotype. “We probably won't exclude E4-positive subjects. But we will be more careful to stratify enrollment and make sure there are enough in each category,” Quinn said.

The MIDAS trial did not obtain participants’ ApoE genotypes, Karin Yurko-Mauro of Martek told Alzforum after her ICAD press briefing. However, Paul Aisen, University of California, San Diego, who heads the ADCS, said, “You have to assume that most people in that trial are E4-negative just based on ApoE’s population distribution and the fact that they were cognitively normal (i.e., did not have AD) at this age.” In the ADCS DHA trial, nearly 60 percent of participants were E4-positive. While ApoE effects are viewed with interest across the field these days, they have not so far panned out in trials designed to follow up on initial Phase 2 ApoE3 differential effects. A recent example is GlaxoSmithKline’s Phase 3 pharmacogenomic trial of the diabetes drug rosiglitazone, which grew out of an unexpected ApoE signal in Phase 2 (see ARF related news story). At this year’s ICAD in Vienna, the company reported that the compound had no impact on any of the ApoE-stratified subgroups (see Doctor's Guide news story). GSK’s Michael Gold noted that rosiglitazone testing for Alzheimer’s had ended (see also page 10 of press release [.pdf]).

Confirming ApoE effects in medical foods could present yet additional complications. For example, if DHA were to show benefit in ApoE4-negative MCI patients, one could imagine the troublesome scenario in which people would need genetic testing to be prescribed a medical food. This could mean that potential consumers who test as E4-negative would be eligible to take something lacking an FDA efficacy statement, while those who come up E4-positive would get hit twice—with news of elevated AD risk, and without another potential treatment option.—Esther Landhuis.

This is Part 1 of a two-part series. See Part 2

Q&A With Sue Anderson, FDA supervisory nutritionist. Questions by Esther Landhuis.

Q: What is a "medical food"?
A: In simple terms, a medical food is a product that is specially made because in no way can a regular diet be modified to meet the specific and distinctive nutritional needs of a disease or condition. It is not a product that is merely provided to sick people, although it may have a beneficial effect. It is a formulated product that contains ingredients shown to be directly linked biochemically or metabolically to the disease or condition in humans.

This website contains the statutory definition that has existed since 1988. There have not been any changes or updates to the definition.

Q: What key features distinguish a medical food from a drug?
A: Basically, intent of use and the provision of nutrition support by a medical food. There are specific requirements for pre-market safety and efficacy studies for new drugs and FDA approval of new drug products that are not required for medical foods.

Q: How is a medical food different from a nutraceutical?
A: There is no regulatory definition or framework for “nutraceuticals.”

Q: When was the "medical foods" designation created? Why?
A: The actual words “medical foods” began to be used in 1975. However, the statutory definition was not created until 1988. The most likely reason for the designation stemmed from the change in regulatory practices for such products. Products for inborn errors of metabolism used to be regulated as drugs, but were reclassified as foods and regulated under food law in 1972 so that the products would not have to undergo the pre-market testing and approval required for drugs.

Q: Can you give examples of where the "medical foods" designation has been granted in other disease areas?
A: No “granting” of a designation or classification is made given that there is no pre-market review or approval for medical foods. [Editor’s note: A review (Morgan and Baggott, 2006) mentions Lofenalac (a powder mixed into formula for babies and children with phenylketonuria) as an early example. Newer medical foods target pancreatic and renal insufficiencies (Ultrase MT and Renax), elevated homocysteine (Folgard RX 2.2), and other conditions including vascular dementia and AD (CerefolinTM)].

Q: Do insurance companies generally reimburse for medical foods? Are they expensive like drugs, or more like a food?
A: FDA does not regulate insurance coverage. The internal policies and practices of individual insurance companies differ. Medical foods are regulated under existing conventional food law and are not drugs. Some insurance companies may provide coverage and some others may not for varying reasons. Medical foods can range in cost from reasonable and affordable to thousands of dollars per year (i.e., medical foods used to manage inborn errors of metabolism).

Q: With no efficacy claim by the FDA, how do consumers know that a medical food will do them any good? Are medical foods required to meet any efficacy standards?
A: The short answer is that consumers don’t know. Manufacturers should have information to substantiate that claims on their labels are truthful and not misleading. However, the claims are not reviewed by FDA.


  1. As noted in this interesting article, I think the approach of supplying alternate forms of bioenergetic substrates to patients with Alzheimer disease is worth further exploration, and future studies must be designed and powered to test a differential APOE response, which we have observed in our own studies of insulin/energy-modulating agents. In the interest of full disclosure, as the article described, I received a small grant from Accera to conduct an acute dosing study of an MCT formulation in 2004; additionally, I also serve as a consultant for Accera, a fact that was not mentioned in the article.

  2. The premise of this article is the notion that companies are using the medical food route as a “fallback” or backup strategy if their drug compound fails in the clinic. As I will discuss below, this premise is flawed. I wish to point out that this was never the intent for Axona (AC-1202). As I stated in the Tangled Neuron interview, Axona was originally intended to be a surrogate for testing our new therapeutic approach (ketone treatment for neuronal hypometabolism) in AD patients so the company could secure venture funding for its drug development platform.
    Based on our research, we found evidence that the dietary addition of ketones can delay and reduce the magnitude of cognitive dysfunction in patients with mild to moderate AD and can be an effective part of the dietary management of the disease. As a result, we concluded that the product could be appropriately marketed as a medical food.
    The company never filed an IND for Axona or intended to develop it as a drug. However, we did perform our clinical studies to pharmaceutical standards with industry and FDA-accepted outcomes for safety and efficacy. These data have been presented at major scientific conferences such as AAN and ICAD, and have been reviewed by internationally acclaimed experts on AD. The clinical data are also incorporated into our 10,000+ page product dossier that supports its positioning as a medical food.
    As a result of our laboratory and clinical studies with Axona during the past seven years we now have two preclinical New Chemical Entity compounds in development as drugs that work through the ketosis therapeutic mechanism.
    In terms of the article’s overall premise, there were some significant inaccuracies. First, NCEs and drug products cannot be marketed as medical food since medical food must meet the general food requirement of having ingredients that are either GRAS or approved food additives. Additionally, as noted in the Q&A with Sue Anderson from the FDA, medical food “is not a product that is merely provided to sick people … It is a formulated product that contains ingredients shown to be directly linked biochemically or metabolically to the disease or condition in humans.” I would also note that Vivimind (aka Alzhemed) is a dietary supplement, not a medical food. Supplements and medical food have distinct definitions and regulatory requirements. We are not aware of any examples of a product in the marketplace that was switched from drug development to medical food.
    While it is true that medical food is not required to report adverse events to FDA, this statement holds true for all food products. Nevertheless, Accera has in place a full medical information and safety system that compiles safety-related information and relays all serious adverse events via MedWatch to the FDA. We believe this is a responsible way to identify and address potential side effects and drug interactions in an elderly patient population that often has many comorbidities.
    Lastly, I wish to point out that the discussion of “what’s in a label” shows a fundamental misunderstanding of the key features that distinguish a medical food from a drug. As Ms. Anderson notes in the Q&A discussion, the product’s intended use and the provision of nutritional support distinguish medical food from a drug. In other words, to be considered a medical food, a product must be intended for the specific dietary management of a disease or condition and must be an integral part of the clinical treatment of patients. It may not, however, state or imply that the product treats, cures, or prevents a disease. Those are drug claims. To imply that the Axona label is insufficient or indicates that the product failed drug trials is simply inaccurate. The Axona label appropriately reflects the product’s intended use as a medical food.

  3. I think it would have been a major advantage to get Ketasyn/AC1202 FDA-approved as a drug. Ketasyn/AC1202 could have then been used by health professionals and prescribed to the right persons. There is a strong rationale in using medium chain triglycerides (MTCs) as a source of ketone bodies to boost brain metabolism. It is likely that certain specific patients in “energy crisis”, such as very old persons for example, could benefit from MCTs. Unfortunately, the use of Ketasyn/AC1202 as a medical food will dilute its true therapeutic benefit.

    In summary, I might be wrong but I think Ketasyn would have had more chance to achieve its full therapeutic potential as a drug than as a medical food.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Washington: Shaking Up AD Treatment with Ketone Bodies
  2. Chicago: Phase 2 News—Therapeutic Breakfast Food?
  3. World Alzheimer’s Report Estimates 35 Million Cases
  4. More ApoE4 Means More Amyloid in Brains of Middle-Aged
  5. Madrid: Highs and Lows of The Insulin Connection
  6. Medical Foods—Food for Thought, But Think Twice

Paper Citations

  1. . Study of the ketogenic agent AC-1202 in mild to moderate Alzheimer's disease: a randomized, double-blind, placebo-controlled, multicenter trial. Nutr Metab (Lond). 2009;6:31. PubMed.
  2. . Medical foods: products for the management of chronic diseases. Nutr Rev. 2006 Nov;64(11):495-501. PubMed.
  3. . Effects of beta-hydroxybutyrate on cognition in memory-impaired adults. Neurobiol Aging. 2004 Mar;25(3):311-4. PubMed.
  4. . Correlations between apolipoprotein E epsilon4 gene dose and brain-imaging measurements of regional hypometabolism. Proc Natl Acad Sci U S A. 2005 Jun 7;102(23):8299-302. PubMed.
  5. . Fibrillar amyloid-beta burden in cognitively normal people at 3 levels of genetic risk for Alzheimer's disease. Proc Natl Acad Sci U S A. 2009 Apr 21;106(16):6820-5. PubMed.

Other Citations

  1. see Q&A below

External Citations

  1. FDA Guidance for Industry
  2. another story
  3. Doctor's Guide news story
  4. page 10 of press release

Further Reading


  1. . Study of the ketogenic agent AC-1202 in mild to moderate Alzheimer's disease: a randomized, double-blind, placebo-controlled, multicenter trial. Nutr Metab (Lond). 2009;6:31. PubMed.
  2. . Medical foods: products for the management of chronic diseases. Nutr Rev. 2006 Nov;64(11):495-501. PubMed.