Agency Response Letter GRAS Notice No. GRN 000041
CFSAN/Office of Premarket Approval
May 17, 2001
Henry Linsert, Jr.
Martek Biosciences Corporation
6480 Dobbin Road
Columbia, Maryland 21045
Re: GRAS Notice No. GRN 000041
Dear Mr. Linsert:
The Food and Drug Administration (FDA) is responding to the notice, dated February 29, 2000, that Hogan and Hartson submitted on behalf of Martek Biosciences Corporation (Martek) in accordance with the agency's proposed regulation, proposed 21 CFR 170.36 (62 FR 18938; April 17, 1997; Substances Generally Recognized as Safe (GRAS)). FDA received the notice on March 1, 2000 and designated it as GRAS Notice No. GRN 000041.
The subject of the notice is DHASCO (docosahexaenoic acid-rich single-cell oil) and ARASCO (arachidonic acid-rich single-cell oil). DHASCO is derived from the microalgal species Crypthecodinium cohnii; ARASCO is derived from the soil fungus Mortierella alpina. In its notice, Martek informs FDA of its view that DHASCO and ARASCO as sources of docosahexaenoic acid (DHA) and arachidonic acid (ARA) are GRAS, through scientific procedures, when added to term infant formulas as described in Table 1.
|Substance||Percent of daily fat||Percent of calories||Mg/kg bw/day|
|ARA||Up to 0.5 percent||Up to 0.25 percent||Up to 30 mg/kg bw/day|
|DHA||Up to 0.5 percent||Up to 0.25 percent||Up to 30 mg/kg bw/day|
|ARA + DHA||Up to 1.0 percent||Up to 0.5 percent||Up to 60 mg/kg bw/day|
|ARASCO||Up to 1.25 percent||Up to 0.625 percent||Up to 75 mg/kg bw/day|
|DHASCO||Up to 1.25 percent||Up to 0.625 percent||Up to 75 mg/kg bw/day|
|ARASCO + DHASCO||Up to 2.5 percent||Up to 1.25 percent||Up to 150 mg/kg bw/day|
*The ratio of DHA to ARA (and of DHASCO to ARASCO) would range from 1:1 to 1:2.
Data and information that Martek presents to support its determination that ARASCO and DHASCO are GRAS for use in infant formula
As part of its notice, Martek includes the report of a panel of individuals (Martek's GRAS Panel) who evaluated the data and information that are the basis for Martek's GRAS determination. Martek considers the members of its GRAS panel to be qualified by scientific training and experience to evaluate the safety of substances added to food.
Martek describes published information about the levels of the fatty acids ARA and DHA in human milk and considers that the presence of ARA and DHA in human milk establishes that these fatty acids are GRAS. Martek cites published information to support its view that ARA and DHA have a role as nutrients. Martek describes the recommendations of several international bodies that these fatty acids be added to preterm infant formulas, term infant formulas, or both, at levels similar to the levels proposed by Martek.
Martek describes published information about the digestion of triglycerides by infants. Martek describes published information relevant to the biodisposition of ARASCO and DHASCO, focusing on a comparison of the biodisposition of ARASCO and DHASCO to that of ARA- and DHA-containing triglycerides in human milk. Martek concludes that the digestion and absorption of DHA from DHASCO, and ARA from ARASCO, would be the same as that of DHA and ARA from triglycerides present in human milk.
Martek describes published information about the identity and characteristic properties of ARASCO and DHASCO, including published information about the source microorganisms for ARASCO and DHASCO and the method of manufacture of ARASCO and DHASCO. Martek provides food grade specifications for ARASCO and DHASCO. Martek describes published information about the composition of sterols in the nonsaponifiable fractions of DHASCO and ARASCO. Martek states that the principal components of the sterol fraction in DHASCO (i.e., 4-methyl sterols) are found in the normal metabolic pathway of cholesterol biosynthesis and have been identified in several common food sources including fish and shellfish. Martek further states that the principal component of the sterol fraction in ARASCO (i.e, desmosterol) also is found in the normal metabolic pathway of cholesterol biosynthesis and is commonly found in several common food sources including animal fat, vegetable oils, and human milk.
Martek describes published and unpublished toxicity studies conducted in rats treated with ARASCO and DHASCO. These studies include acute studies, 28-day studies, a 63-day study, 90-day studies, and developmental and reproductive toxicity studies. Martek concludes that the relatively large number of toxicity studies provides a degree of redundancy in the standard toxicological assessments.
In Martek's view, ARASCO and DHASCO pose unique testing problems because it can be difficult to distinguish whether an observed effect related to the test material is a normal physiological response to the high dietary load of that particular macronutrient, a dietary deficiency that is related to presence of a large amount of the test material in the diet, or a toxicological effect.(1). Martek acknowledges that some of the available animal toxicity studies have reported statistically significant effects at the highest doses tested, including an increase in relative liver weights, an increase in relative spleen weights, and a change in some blood chemistries. For each of these reported effects, Martek explains the reasons for its conclusion that the reported effects are not adverse toxicological findings that are related to ARASCO and DHASCO as sources of ARA and DHA. Martek concludes that the No Observed Effect Level (NOEL) in each study should be the highest dose tested.
Martek also acknowledges that some studies have reported changes attributed to possible effects on the kidney. Martek presents an analysis of these studies and concludes that there are no consistent treatment related effects of DHASCO and ARASCO, either alone or in combination, with respect to kidney weight, kidney function or kidney histopathology. Martek discusses a published subchronic rat study with an in utero phase and a published developmental toxicology study and concludes that there is no evidence of any treatment-related adverse reproductive effects in rats treated with DHASCO and ARASCO.
Martek cites 14 published clinical intervention trials involving 1,500 term or preterm infants. In all, more than 700 of the term or preterm infants enrolled in these studies consumed formulas supplemented with DHASCO or ARASCO oils. Martek reports that none of the studies conducted with ARASCO and DHASCO as sources of ARA and DHA demonstrated any adverse effect on infant growth.
Martek notes that some studies reported that infants who consumed term and preterm infant formula supplemented with fish oil as a source of DHA exhibited reduced growth and a decline in measures of ARA status. To address these reports, Martek notes that fish oil contains eicosapentaenoic acid (EPA), which is not present in DHASCO and which is known to be an antagonist of ARA metabolism. In Martek's view, the reduced growth seen in these studies could be related to (1) the presence of EPA, which is present in fish oils; (2) the lack of a source of ARA in the formula supplemented with fish oil; or (3) a combination of the presence of EPA and the lack of ARA. Martek contrasts the reports of negative effects on growth in infants who consumed formulas supplemented with fish oil to the reports of studies conducted with DHASCO as a source of DHA, which have not reported negative effects on growth.
Martek presents information about DHASCO- and ARASCO-supplemented infant formulas that have been introduced to the international marketplace during the past 5 years. According to Martek, more than 100,000 infants have consumed a formula that is marketed for low birth weight infants and is administered under a physician's strict supervision, with no reported adverse effects associated with the consumption of the formula.
Martek discusses a published clinical study that reported more deaths attributed to Sudden Infant Death Syndrome (SIDS) in infants who consumed a formula supplemented with fish oil compared to infants in the control group. According to Martek, an independent safety committee, which included specialists with expertise in SIDS, carefully reviewed each case and concluded that none of the SIDS deaths was related to dietary treatment. In support of the opinion expressed by the independent safety committee, Martek notes that more than 30 other clinical trials have been conducted using formulas supplemented with DHA and ARA from various sources, and that none of these other trials reported any similar observation. Martek also notes that there have been no reports of an increase in SIDS-related deaths in countries where formulas supplemented with a source of DHA and ARA have been used for several years. Martek also notes that there have been no concerns reported by physicians worldwide who have monitored the administration of formulas supplemented with DHASCO and ARASCO to low birth-weight infants.
FDA's evaluation of the data and information in Martek's notice
On January 29, 2001, at Martek's request, Martek met with representatives of the Center for Food Safety and Applied Nutrition (CFSAN). In a letter dated January 31, 2001, you reiterated a concern that you expressed on January 29 about a delay in the agency's response to your GRAS notice. In your January 31 letter, you outlined your understanding of a commitment made by CFSAN at the January 29 meeting to define, as soon as possible, any specific unresolved scientific questions related to this notice.
In a letter dated March 1, 2001, CFSAN explained that the delay in its response to your GRAS notice was related to broader issues than those related to your notice and described specific scientific concerns related to the use of infant formulas containing long chain polyunsaturated fatty acids (LCPUFAs) such as the fatty acids ARA and DHA. As we explained in our March 1 letter, some publications and panel reports, which considered multiple sources of ARA and DHA (e.g, sources such as fish oil and egg phospholipid), have questioned the scientific adequacy of data to support the use of LCPUFAs in infant formula. In addition, some studies have reported unexpected deaths among infants who consumed formula supplemented with LCPUFAs. These unexpected deaths were attributed to Sudden Infant Death Syndrome (SIDS), sepsis or necrotizing enterocolitis. Also, some studies have reported adverse events and other morbidities including diarrhea, flatulence, jaundice, and apnea in infants fed LCPUFAs. In addition, CFSAN noted that your notice had not accounted for the fact that the bioactive fatty acids ARA and DHA when consumed in mature human milk are part of a complex matrix that includes, for example, linoleic acid, alpha-linolenic acid, and other polyunsaturated fatty acids and that important physiologic considerations relative to the matrix are not accounted for by the simple addition of LCPUFAs to infant formula.
In our letter dated March 1, 2001, we informed you that CFSAN intended to convene a group of experts to address the broader scientific issues raised by your specific determination that the ingredients DHASCO and ARASCO are GRAS sources of DHA and ARA for use in infant formula.
In a letter dated March 23, 2001, Martek expressed its view that the reports and publications that questioned the scientific adequacy of data to support the use of LCPUFAs in infant formula recommended additional studies, which are now available. Martek also noted that studies conducted in term and preterm infants with balanced addition of ARASCO and DHASCO as sources of both bioactive fatty acids showed no difference in deaths between treatment and control groups. In Martek's view, the results of clinical studies conducted with infant formulas manufactured by three different firms are the most direct and conclusive evidence that there are no concerns related to physiologic considerations associated with the overall physiological matrix of infant formula compared to human milk.
In its March 23 letter, Martek expressed its view that some of the concerns described in CFSAN's March 1 letter are hypothetical and that convening a group of scientific experts to answer such hypothetical concerns would not be productive.
Based on all of the information provided by Martek, as well as other information available to FDA, the agency has no questions at this time regarding Martek's conclusion that ARASCO and DHASCO are GRAS sources of ARA and DHA under the intended conditions of use - i.e., when added to infant formulas intended for consumption by healthy term infants at a level of up to 1.25 percent each of total dietary fat and at a ratio of DHA to ARA of 1:1 to 1:2. The agency has not, however, made its own determination regarding the GRAS status of the subject use of ARASCO and DHASCO. As always, it is the continuing responsibility of Martek to ensure that food ingredients that the firm markets are safe. As discussed in detail below, it is the responsibility of an end user of ingredients that Martek markets for use in infant formula to ensure that an infant formula that contains such ingredients is otherwise in compliance with all applicable legal and regulatory requirements.
As noted above, in its March 23 letter, Martek expressed its view that convening a panel of experts to discuss the broad scientific issues raised by its notice would not be productive. We disagree with that view. Rather, we have initiated a discussion with an authoritative body of scientific experts to address those broader issues, such as, for example, the addition of bioactive ingredients to an infant formula matrix, which is different from the physiological matrix of mature human milk.
As you are aware, under section 412 of the Federal Food, Drug, and Cosmetic Act (the FFDCA), a manufacturer of a new infant formula must make a submission to FDA, providing required assurances about the formula, at least 90 days before the formula is marketed. We realize that Martek is a supplier of an ingredient that would be used in infant formula, rather than a manufacturer of infant formula. Thus, we would not expect Martek to be making the submission required by section 412 about an intent to market a new infant formula that contains ARASCO and DHASCO. However, you should be aware that FDA's response to your GRAS notice does not alleviate the responsibility of any infant formula manufacturer who intends to market an infant formula that contains ARASCO and DHASCO to make the submission required by section 412.
As noted in our March 1 letter, the specific use of the ingredients DHASCO and ARASCO in infant formula raises broad scientific issues about the use of bioactive ingredients such as LCPUFAs in infant formula. In addition, it is FDA's view that any evaluation that a use of a food ingredient is safe is a time-dependent judgment that is based on general scientific knowledge as well as specific data and information about the ingredient. For these reasons, FDA would expect any infant formula manufacturer who lawfully markets infant formula containing ARASCO and DHASCO to monitor, through scientific studies and rigorous post-market surveillance, infants who consume such a formula. We also would expect regular reports of such studies and post-market surveillance. Because the use of ARASCO and DHASCO in infant formula would be based on the GRAS provision of the FFDCA, we also would expect that these reports would not be considered to be confidential so that the broader scientific community can contribute to this continuing evaluation. In light of these expectations, we strongly encourage any manufacturer of infant formula who is considering using Martek's ARASCO and DHASCO in infant formula to consult with FDA's Office of Nutritional Products, Labeling, and Dietary Supplements (ONPLDS) before making the required premarket submission. Moreover, we strongly encourage any such manufacturer to consult with ONPLDS before initiating premarket clinical studies, and FDA would plan to engage manufacturers concerning any needed postmarket clinical studies or postmarket surveillance.
In accordance with proposed 21 CFR 170.36(f), a copy of the text of this letter, as well as a copy of the information in your notice that conforms to the information in proposed 21 CFR 170.36(c)(1), is available for public review and copying on the Office of Premarket Approval's homepage on the Internet (at http://www.cfsan.fda.gov/~lrd/foodadd.html).
Alan M. Rulis, Ph.D.
Office of Premarket Approval
Center for Food Safety and Applied Nutrition
Christine J. Lewis, Ph.D., R.D.
Office of Nutritional Products, Labeling, and Dietary Supplements
Center for Food Safety and Applied Nutrition
(1)As an analogy, Martek presents a scenario of elevated serum cholesterol in a human who changes a diet from one with 25 percent of calories from fat to one with 45 percent of calories from fat. In this analogy, Martek notes that the elevation in serum cholesterol ordinarily is viewed as a normal physiological response to increased dietary fat rather than as a toxicological effect of the particular fat used.