Summary¹

Infant formula is consumed by two-thirds of infants in the United States and is the sole source of nutrition for many of these infants during the first year of life, making its safety and quality particularly critical. Infants are a vulnerable population experiencing rapid and critical growth and development. The U.S. Food and Drug Administration (FDA) oversees the safety and adequacy of infant formula in the United States. While most regulations that apply to food also apply to infant formula, it has additional requirements specific to infant formula, which appear in section 201(z) of the Food, Drug & Cosmetic Act. These additional specifications include the quality factor requirements of (1) the assessment of the biological quality of protein, generally through the protein efficiency ratio method and (2) the assessment of normal physical growth of the infant through growth monitoring studies. The legislative and regulatory actions on quality factors have evolved over time (see Table S-1).

The 2023 Consolidated Appropriations Act (P.L. 117-328) directed FDA to enter into an agreement with the National Academies of Sciences, Engineering, and Medicine (the National Academies) “to examine and report on challenges in supply, market competition, and regulation of infant formula in the United States.” The act also included other provisions that encourage FDA to consider approaches to harmonizing its regulatory requirements with those used in other countries, which may

___________________

¹ This summary does not include references. Citations for findings presented in the summary appear in the subsequent chapters of the report.

TABLE S-1 Timeline of Legislative and Regulatory Actions on Quality Factors

NOTES: CFR = Code of Federal Regulations; FR = Federal Register; IFA = Infant Formula Act; IFR = Interim Final Rule; P.L. = Public Law; QF = quality factors; STAT = statute. In 2002 and 2003, the U.S. Food and Drug Administration also held Food Advisory Committee meetings, 67 FR 12571, March 19, 2002; 67 FR 63933, October 16, 2002; and 68 FR 8299, February 20, 2003, and reopened the comment period to solicit comments on quality factors, among other things, 66 FR 20589, April 28, 2003.

SOURCES: P.L. 96-359—September 26, 1980; P.L. 100 STAT. 3207-38; October 27, 1986; 61 FR 36154, September 23, 1996; 79 FR 7934, February 10, 2014; 79 FR 33037, June 10, 2014.

be useful, particularly during shortages of domestic production of infant formula, as in 2022. FDA requested the National Academies to examine and report on the state of the science regarding methodologies for assessing the biological quality of protein in an infant formula and the ability of an infant formula to support normal physical growth. The report was to include consideration of study designs commonly used in other global regions (see Box S-1).

The National Academies convened a committee with expertise spanning public health policy; infant health and nutrition; composition of

infant formulas, both global and domestic; protein chemistry, biological efficiency, and quality; design and conduct of in vivo and in vitro studies; overall methodologies to assess protein quality, including animal models; design and conduct of clinical studies; assessment of human growth; and standards for good laboratory practices, clinical practices, and protection of human participants.

THE COMMITTEE’S APPROACH

The committee interpreted the statement of task as a request to understand and describe the state of the science on the biological quality of protein in an infant formula and the ability of an infant formula to support normal physical growth. The committee denotes weight gain as the conventionally used primary outcome variable to assess normal physical growth. The committee commissioned the Academy of Nutrition and Dietetics (AND) to carry out two scoping reviews, one for each quality factor requirement. The aim included (1) examining and reporting on the methods or study design, including strengths and weaknesses; (2) assessing whether the methods have been validated; and (3) examining whether the methods vary globally. The committee developed two

sets of questions to guide the scoping reviews (see Box S-2). With guidance from the committee, AND developed a scoping review protocol, which included a search strategy and methods for screening and data extraction. The committee also held public information-gathering sessions, assessed public documentation provided by the sponsor (FDA) and published by other authoritative bodies, and carried out narrative reviews for any additional needed information related to the task. The committee provided conclusions and recommendations relevant to its task, although not all conclusions led to a recommendation.

Protein Quality Scoping Review

The AND scoping review on the topic of protein quality yielded 1,806 unique articles in the database searches. The full texts of 145 articles were reviewed against eligibility criteria for inclusion, and 41 studies were included in the scoping review (Table 2-1): animal, in vitro digestibility, and validity/reliability studies. Thirty-five of the 41 studies examined cow’s milk–based formula, and five investigated multiple types of formula, including that with partially hydrolyzed protein. Six studies examined plant-based formulas, and five examined goat’s milk–based formulas.

Normal Physical Growth Scoping Review

The AND scoping review on the topic of normal physical growth yielded 4,927 unique articles in the database searches, of which 235 met the eligibility criteria, and 143 were included in the scoping review (Table 2-2): 16 were systematic reviews, and 125 focused on 115 primary studies. The greatest percentage of studies were conducted in the United States, followed by Italy, Spain, Germany, and China. Most studies were funded by infant formula companies, followed by government. The study sample sizes were 100–500 participants, with an equal distribution of boys and girls in most studies that reported this information. The type of formula was primarily cow’s milk based, with a few using protein from plants, soy, or goat’s milk. The modification for the experimental formulas included the amount and type of protein, hydrolysis of protein, amino acid (AA) modification, fat amount, fat type, presence of micronutrients, removal of lactose, and inclusion of prebiotics and/or probiotics.

PROTEIN QUALITY

FDA proposed a rule that set out nutrient specifications for infant formula (49 FR 14396, April 11, 1984) which included that the biological quality of protein be not less than 70 percent of casein. This requirement was finalized in regulation in 1985 (21 CFR § 107.100). In 1996, FDA proposed to establish a quality factor for protein in infant formula, which was finalized in 2014 (79 FR 33037, June 10, 2014). Infant formula is shown to meet this quality factor when the manufacturer demonstrates that it meets the sufficient biological quantity of protein using an appropriate modification of the protein efficiency ratio (PER). The final rule states that an alternate method to PER could be used if the manufacturer assures that it is based on sound scientific principles. In 2023, FDA proposed Draft Guidance for Industry on conducting the PER bioassay, enumerating dietary modifications to more closely match the test and control diets. However, as of December 2024, this draft guidance had not been finalized.

Protein Efficiency Ratio

The PER assay assesses the quality of a protein by using a rodent model to measure weight gains after consuming the test protein (that is, a test infant formula containing the protein) and comparing the result to weight gain of rats fed a control diet with casein as the protein source. PER measures the biological quality of the protein in a processed infant formula product.

Infant formula processing may generate reaction products that could render nutrients, including AAs, unavailable to the body. FDA has main-

TABLE S-2 Decision Matrix for the Assessment of Protein Quality in Infant Formula

NOTES: DIAAS = Digestible Indispensable Amino Acid Score; HMAA = human milk amino acid pattern.

^a Major changes (21 CFR § 106.3) that involve protein include any infant formula produced by a manufacturer that is entering the U.S. market or having a significant revision, addition, or substitution of a macronutrient (i.e., protein, fat, or carbohydrate) with which the manufacturer has not had previous experience. The committee considered new protein to include (a) a new species of protein, (b) a new mixture or large quantitative difference of existing proteins (American Academy of Pediatrics described as 20 percent), or a quantitative level at or below the minimum in regulation (where Codex and European regulations trigger a clinical evaluation).

^b Major changes that involve other ingredients include any infant formula having a significant revision, addition, or substitution of a macronutrient (i.e., protein, fat, or carbohydrate) with which the manufacturer has not had previous experience and any infant formula manufactured containing a new constituent not listed in section 412(i) of the Federal Food, Drug, and Cosmetic Act (21 U.S.C. 350a(i)), such as taurine or L-carnitine.

^c Major processing change is defined as any infant formula powder processed and distributed by a manufacturer that previously only produced liquids (or vice versa); any infant formula manufactured on a new processing line or in a new plant; any infant formula processed by a manufacturer on new equipment that uses a new technology or principle (e.g., from terminal sterilization to aseptic processing); or an infant formula that has a fundamental change in the type of packaging used (e.g., changing from metal cans to plastic pouches).

^d DIAAS = Digestible Indispensable Amino Acid Score; the preferred in vivo method for assessing digestion and accessibility of the products of digestion. HMAA = human milk amino acid pattern; provides evidence of balance among and amount of indispensable amino acids to meet requirements. Animal growth provides evidence of use, and piglets are the preferred model.

tained an emphasis on the bioavailability of (AAs from) protein (the degree to which a nutrient is absorbed or otherwise becomes available to the body), as distinct from protein quality. Other definitions of protein quality involve the age-appropriate AA score of a candidate protein relative to a reference, corrected for digestibility (historically measured by the difference between dietary and fecal protein).

PER has raised criticism from groups such as the World Health Organization (WHO), Food and Agriculture Organization (FAO), and the Life Sciences Research Office (LSRO). The method is not globally recognized by regulatory bodies, frequently used in research studies, or generally commercially available. The use of animal testing for regulatory purposes has been challenged on physiological, biological relevancy, and ethical standpoints. Regulatory bodies outside of North America no longer use PER to determine protein quality of infant formula. Instead, the global standard is the comparison to the AA composition of human milk.

Conclusion 1: The committee concluded that the protein efficiency ratio (PER) is not the preferred method for assessing protein quality of infant formula.

Recommendation 1: The Food and Drug Administration should not use the protein efficiency ratio (PER) as the method for establishing the biological quality of protein of new infant formulas and should reconsider the need for the existing draft guidance on PER.

Amino Acid Composition Compared to Human Milk

The European Food Safety Authority (EFSA) and Codex Alimentarius (Codex) concluded that the composition of human milk from a healthy, well-nourished mother can provide guidance for the composition of infant formula and established the requirement that formula contain the level of every indispensable² (essential) AA, and conditionally indispensable amino acid (cysteine and tyrosine) in human milk. Recent evidence shows that the composition of individual AAs on a mg/g N from human milk is relatively consistent over the first 6 months, confirming the validity of using that pattern of AAs as a reference for infant formula composition.

Conclusion 2: Based on its assessment of the evidence related to the lack of use of PER to assess protein quality and review of relevant regulations in countries outside of the United States, the preferred method of assess-

___________________

² The term indispensable is equivalent to essential amino acid.

ment of protein quality in infant formula is to match the indispensable and conditionally indispensable amino acid composition of human milk. An advantage of this approach is that a published amino acid scoring pattern is available in Codex, the European Food Safety Authority, the Food Standards Australia New Zealand (FSANZ), and the European Union which would contribute to international regulatory harmonization.

Recommendation 2: The Food and Drug Administration should adopt the human milk amino acid pattern as the reference pattern to assess the protein quality of infant formula.

Considerations to Alternative Approaches to Measuring Protein Quality

Alternative approaches identified by the committee include the assessment of protein digestibility and availability, human studies, and in vitro approaches.

Assessment of Protein Digestibility and Availability

An alternative approach to measuring protein quality is to assess the digestibility and availability using animal bioassays. Swine and rodent models have been used to estimate protein digestibility in humans, including infants, and assessment has used fecal and ileal digestibility. The method to assess fecal digestibility is the Protein Digestibility–Corrected Amino Acid Score (PDCAAS), which evaluates protein quality based on AA requirements and digestibility. The scoping review identified two articles that used animal models to estimate PDCAAS in infant formulas. The method to assess ileal digestibility is the Digestible Indispensable Amino Acid Score (DIAAS), which requires direct measurement of the digestibility of each indispensable AA in the protein at the end of the small intestine. The scoping review identified two articles that used animal models to estimate DIAAS of infant formulas.

Human Studies for the Assessment of Protein Quality

The direct measurement of protein quality in infant formula using human studies is difficult due to the complexities of obtaining ileal digesta following a test protein intake. In humans, mostly adults, direct (oro-ileal digestibility) and indirect (dual isotope tracer method and indicator AA oxidation) methods have been developed to measure protein and AA digestibility. No studies were identified by the scoping review that used these methods for infant formulas.

In Vitro Approaches

Another alternative approach to measuring protein quality is in vitro models of PER estimation, which includes both static and dynamic methods.

Static in vitro assays treat suspensions of food with a mix of digestive enzymes. These assays are research tools to study food structure, digestibility, and nutrient bioavailability and provide protein digestibility coefficients. They include pH drop, pH-stat, INFOGEST 2.0, and commercially available kits. The scoping review revealed eight studies using these models to examine the effect of infant formula composition and processing on digestion characteristics. The strengths of these methods include ease of implementation across different lab settings, high throughput, and lower cost compared to in vivo methods. However, such models have not been validated for individual AA digestibility.

Dynamic digestibility assays are designed to simulate gastrointestinal digestion phases and nutrient bioaccessibility using sophisticated, computer-controlled, temperature-regulated digestion chambers. The scoping review yielded six studies that used a two-step semi-dynamic in vitro model designed to mimic infants’ digestion and five studies that used a dynamic in vitro model. The two-step semi dynamic in vitro model represents a better simulation of physiological conditions than static in vitro methods. The strengths of this approach include its ability to mimic gradual changes of pH and enzyme activity and fluids, and therefore allow performance of kinetic studies. The approach can be readily tailored to specific life stages and allows for integrative study of nutrient digestion. Challenges with this approach include the various conformations that are available with different characteristics, e.g., to mimic gastric and peristaltic movements or fluid additions, that make harmonization between laboratories more difficult. The high cost of system acquisition, service contracts, and operational costs, as well as limited suppliers and sample throughput are additional challenges.

Conclusion 3: Changes in protein composition or processing of infant formula may require evidence of digestion, absorption, and utilization of the amino acids in the protein source of the formula. Both in vivo and in vitro alternative approaches to the PER are available to demonstrate digestion. Specific protein quality measures targeted for application in different scenarios of manufacturing processes and formulation changes may be warranted (see Table S-2).

Recommendation 3: The Food and Drug Administration should provide guidance to manufacturers of infant formula when

there is need for evidence of digestibility and bioavailability and on acceptable methods of assuring protein quality that reflect the types of changes in the composition or processing of the formula.

NORMAL PHYSICAL GROWTH

The origin of “normal physical growth” as a quality factor was explained in FDA’s proposed rule in 1996 (61 FR 36154, September 23, 1996). The quality factor reflected the Committee on Nutrition of the American Academy of Pediatrics recommendation to FDA that physical measures of growth, such as weight gain, are valuable clinical evaluations of an infant formula and widely accepted markers of the ability to utilize and digest infant formula nutrients. FDA ultimately determined that the measurement of weight gain, length, head circumference, and formula intake in a growth monitoring study (GMS) is needed for a new infant formula.

The GMS requirements in FDA regulations specify that the new infant formula must support normal physical growth in infants when fed as a sole source of nutrition by conducting, in accordance with good clinical practice, an adequate and well-controlled growth monitoring study of the infant formula.

The terms “adequate” and “well-controlled” are not defined in the regulations or any accompanying publicly available FDA guidance document, nor are details on suitable trial design criteria provided. An exemption is considered from the requirements of a GMS when an alternative method or study design based on sound scientific principles is available to show that the formula supports normal physical growth in infants when fed as the sole source of nutrition, but few examples of such situations were available to the committee.

In brief, FDA regulations (21 CFR § 106.96(b)) state that the GMS must:

• be not less than 15 weeks in duration;
• enroll infants no more than 2 weeks old at time of entry into the study;
• measure body weight, length, and head circumference and calculate average daily weight increment and average daily recumbent length increment at the beginning and end of the study and four intermediate time points with three measurements within the first 4 weeks and three measurements monthly thereafter;
• have a concurrent control group; and

• compare individual subjects’ weight for age, length for age, and head circumference for age, and weight for length in the test and control groups to the 2009 CDC growth charts.

Current Regulatory Requirements for Growth Monitoring Studies

The committee had little information from sources provided to it or studies identified in the scoping review about how FDA used the data required by paragraph 21 CFR § 106.96(b) to determine if infants showed normal physical growth. Guidelines published by other regulatory bodies and agencies and details of study design requirements are provided in other jurisdictions, such as Health Canada and EFSA.

While inclusion of a reference breastfed group is not a requirement of 21 CFR § 106.96, the committee found that, although it is not essential for assessing normal physical growth, it is of value for assessing the physiologic effects of novel ingredients added to infant formula and biological or metabolic outcomes beyond growth. In GMSs, the committee found that a comparison of the group means to specified growth standards (i.e., CDC growth charts for anthropometric z-scores) is important to demonstrate normal physical growth. Finally, the standardization of anthropometric data collection methodology is critical for quality data collection.

Conclusion 4: The committee was unable to identify empirical evidence that supported or refuted the appropriateness of the current requirements as specified in 21 CFR § 106.96(b) and 106.121, or that would support the suitability of alternative designs for evaluation of normal physical growth. It is possible that deviations from these regulations could be scientifically justified and therefore acceptable to the FDA as alternate methods or designs; as well, acceptance of alternative designs may have the additional benefit of contributing to international harmonization of regulatory statutes. Specifically, the committee was not able to establish information to support the necessity of any specific number of anthropometric data collection times nor schedule of those times during the course of a growth monitoring study. This was also the case for supporting a mandatory enrollment age of less than 2 weeks of age or a 15-week trial duration. Other mandated data, such as formula intake and anthropometric measurements other than weight, were not clearly identified as being needed in all growth studies. Decentralized designs whereby anthropometric data collection is performed by parents/caregivers or health professionals who are not study personnel may be an alternative method if suitable efforts are made to ensure that non-study personnel have met minimum standards of precision and accuracy in the conduct of the measurements.

Recommendation 4: The Food and Drug Administration (FDA) should publish a single guidance document that describes: (1) the preferred design features of a growth monitoring protocol and explains how FDA uses required information in its evaluation that a formula supports normal physical growth, and the conditions under which alternative designs may be acceptable to FDA; and (2) guidance that outlines the conditions under which a growth monitoring study is needed. That guidance should take into account (1) whether the change in infant formula could reasonably affect growth, (2) if a new ingredient is normally found in human milk, (3) the extent to which prior studies have examined the effect of a new ingredient on growth, and (4) information about the effects of addition of the ingredient on the level of or bioavailability of a nutrient, whose deficiency over the course of the study would be manifested in reduced growth.

Conclusion 5: Current FDA regulations do not clearly define the nature of a major change to an infant formula that requires a growth monitoring study (GMS) other than the requirement that all ingredients in infant formula must be shown to be safe and suitable for their use in formula (21 CFR § 106.40). Guidance is lacking on the specific type of evidence required under various conditions of change in infant formula such as potential effects of a change in formula composition or processing on infant growth; if a new ingredient is normally found in human milk; the extent to which prior studies have examined the effect of a new ingredient on growth; and information about the effects of addition of the ingredient on the level of or bioavailability of a nutrient whose deficiency over the course of the study would be manifested in reduced growth.

Conclusion 6: Weight gain is conventionally the primary outcome variable used to assess normal physical growth. However, there is a lack of strong evidence to support the commonly implemented, but not described in regulation, usage of a 3g/day maximum allowable difference in weight gain between test and control groups. Furthermore, there is no statistical justification to prioritize the analysis of average weight gain over the analysis of endline weight as the primary outcome in the context of a randomized controlled trial. Estimation of differences in endline mean z-scores when comparing test and control groups, or when comparing any group to the WHO/CDC growth standard, is a reasonable alternative to comparisons based on differences in weight gain. The establishment of a standard acceptable non-inferiority or equivalence margin (e.g., ±0.5 z-scores) could facilitate international harmonization of regulatory requirements.

Alternative Approaches

Alternative approaches to GMS anthropometric data collection include variations in the number of data collection points, location, individuals collecting the data, and discontinuation of some measurements. Each variation has potential benefits and limitations.

Anthropometric measures, predominantly weight velocity, body mass index (BMI), weight z-scores or weight-for-age or length z-scores are used most often in GMSs, as such measures are relatively easy and inexpensive to assess or calculate and efficient for larger sample sizes. Limitations include interobserver variations and the difficulty of positioning of infants on length boards due to movement. A drawback with anthropometric measures is that they do not directly measure fat or lean mass or other aspects of normal physical growth and development.

FDA has long recognized that the many types of changes to the composition or processing of an infant formula do not require the same type of supportive evidence to establish its safety and nutritional suitability. Given that no GMS is required for many major changes, it is plausible that alternative methods to a randomized controlled trial (RCT) might satisfactorily document normal physical growth when a GMS is needed. The committee recognizes that there is no empirical evidence to support a risk-based approach; however, the evidence from its literature scan suggests that physical growth outcomes may not be responsive to modifications of a formula that, by regulation, satisfies all known nutrient requirements for infant growth.

Conclusion 7: Conducting a research study in which a new formula is compared to an existing approved one, referred to by the term “concurrent control” as used in 21 CFR § 106.96(b)(5), is conventionally interpreted by investigators and FDA to mean the need for a randomized controlled trial (RCT). An RCT may be needed to demonstrate the absence of a negative effect on growth of infant formula-fed infants because of a change in formulation or processing of an infant formula, However, an RCT may not be needed under certain conditions; and suitable data could be generated in a single-arm study in which the growth of infants receiving the test formula is compared to the WHO/CDC growth standard.

FUTURE RESEARCH ON QUALITY FACTORS

Protein Quality

Adopting the human milk amino acid pattern (HMAA) as the reference pattern to assess protein quality in infant formula and using different

methods to assess protein quality that depend on the type of change constituting a “new” infant formula each raises research needs.

The adoption of the HMAA as the reference should be refined by research, preferably on a global basis. The contribution that geographical region, season, ethnicity, age, nutritional status and behavioral habits may or may not have on the pattern of indispensable and conditionally indispensable AAs needs to be assessed. Units of measure and conversion factors (mg or µmole AA per g nitrogen, mg/100kcal) should be standardized to enable comparison among different studies, as should the methodology for the AA analysis of human milk. For example, understanding the influence of different processing times on protein hydrolysis and reference values for indispensable and conditionally indispensable AAs is needed.

While the committee accepts that digestibility of protein/AAs in human milk need not be of concern to satisfy the infant’s AA requirements, standardized procedures for assessing digestibility of protein sources in infant formula are a necessity, particularly when new sources are introduced.

Over the past 25 years, the young, growing pig has emerged as the premier pre-clinical model for the study of pediatric nutrition due to its comparable anatomy, physiology, and metabolism to the human infant and its adaptability to artificial rearing systems. To ensure rigor and reproducibility, standardized procedures for conducting digestibility and growth monitoring studies in piglets should be established. A standard set of metadata should be required including the selection of breed, age, and sex of the animals, housing conditions, mode and timing of feeding, length of study and composition of the control formula. The committee reviewed several in vitro digestibility assays used to assess protein and AA digestibility; as of December 2024, no method has been sufficiently tested and validated for infant formulas. There is a need to standardize in vitro digestibility methods and validate them against in vivo methods.

The formation of adducts/Maillard reaction products when milk protein components react with lactose is well known. Research is needed to understand how reaction products influence digestibility of indispensable and conditionally indispensable AAs. Furthermore, infant formula processing might impact bioavailability of AAs. The impact on protein quality of protein sources needs to be quantified to ensure that “availability” of AAs is not compromised in infant formulas.

Growth Monitoring Studies

With limited information available on “alternative study designs” that have been considered to be acceptable (or not) to FDA to conduct and report an adequate GMS, an evidence gap was identified on what

could qualify. To determine alternative study design, information on GMS submissions and an understanding of regulatory requirement interpretation and implementation would be needed to assess how FDA interprets design features implemented in GMSs. Published guidance on alternative study designs would support regulatory adherence by all formula manufacturers with and without experience with the FDA process.

While body composition techniques beyond routine anthropometry are not currently used by the FDA in its evaluation of infant formula, these techniques are an active area of research. Assessment of body composition using air-displacement plethysmography, dual-energy X-ray absorptiometry, or multisite skinfold thickness is now more widely available, and normal reference standards have been published. In the few studies identified in the scoping review, some showed significant differences in body fat and lean mass in response to feeding different formulas and between formula versus human milk. Population-based reference data for body composition measures of percentage fat or lean mass or fat or lean mass index are being collected and should be considered in developing measures of quality factor assessment of infant formulas. Early-onset alterations in body composition may have long-term health consequences, and future research should be done on measures of total body fat and lean mass as quality factors in assessing efficacy of experimental formulas.

In the scoping review, 17 of the 143 included studies used biomarkers as measures in RCTs designed to investigate an experimental formula with specific modifications in the content or composition (see Appendix E). Biomarkers can be useful in determining the efficacy of an infant formula in supporting additional measures of growth and development other than normal physical growth. They may also be useful in predicting long-term health outcomes and substantiating specific label claims. The scoping review was not conducted in a way to determine if biomarkers could be used as a complementary measure to anthropometry measures and did not capture any articles that support the validity of biomarkers that would serve a role beyond anthropometry.

It would be valuable to develop sensitive and specific biomarkers that can be correlated with current and long-term growth and health outcomes of infants, especially in evaluating the safety, potential side effects, or long-term impact on growth, metabolic, immune, or developmental outcomes of a new formula.

This page intentionally left blank.