The EFSA Journal (2005) 280, 1-16

Transcription

1 The EFSA Journal (2005) 280, 1-16 Opinion of the Scientific Panel on Dietetic Products, Nutrition and Allergies on a request from the Commission related to the safety and suitability for particular nutritional use by infants of formula based on whey protein partial hydrolysates with a protein content of at least 1.9 g protein/100 kcal (Request No. EFSA-Q ) (adopted on 5 October 2005) SUMMARY The Panel has been asked to provide a scientific opinion on the suitability for the particular nutritional requirements of infants and young children of an infant formulae and follow-on formulae based on protein partial hydrolysates with a protein content of 1.9 g/100 kcal. The conclusions of the Panel are as follows: The formula is as suitable to satisfy the particular nutritional requirements of young infants and as safe as a formula based on hydrolysed whey protein with a higher protein content when fed ad libitum. This conclusion is primarily based upon the growth data of term-infants who were studied between the ages of 8 to 112 days. The nitrogen balance data and the protein status parameters are considered to be supportive of this conclusion. The conclusion on the suitability of the formula with a crude protein content of 1.9 g/100 kcal is specific for this formulation and the protein source it contains. While no data on follow-on formula based on hydrolysed whey protein with a protein content of 1.9 g/100 kcal have been submitted, the Panel considers that a formula with this protein formulation is suitable for use in older infants in conjunction with complementary foods. Any infant formula with a similar crude protein content but differing in protein quality, protein sources or protein processing needs to be tested in a clinical study and its safety should be documented and assessed. Recommendations on how to assess the protein quality, sufficiency and suitability in infant formulae have been published. They include a clinical trial of three to four months duration in a group of healthy term infants with a number to provide sufficient statistical power in comparison to an existing infant formula and/or a reference group of breast-fed infants, and starting preferably at or near birth. Data on growth, protein status and on formula intake should be collected. KEY WORDS Infant formula, infant nutrition, protein requirement, partially hydrolysed whey protein, lowprotein formula, growth. Page 1 of 16

2 BACKGROUND Commission Directive 91/321/EEC on infant formulae and follow-on formulae specifies the essential composition of these products. The Community legislation specifies that infant formulae made with protein partial hydrolysates must have a minimum protein content of 2.25 g/100 kcal. The Commission has received a request from a company for the evaluation of a formula based on whey protein partial hydrolysates in view of allowing the placing on the market of such products that have a lower protein content than specified in the existing legislation. The company has developed a formula based on whey protein partial hydrolysates with a protein content of 1.9 g/100 kcal. TERMS OF REFERENCE In accordance with Article 29 (1) (a) of Regulation (EC) No 178/2002, the European Commission asks the European Food Safety Authority to provide a scientific opinion on the suitability of infant formulae and follow-on formulae based on whey protein partial hydrolysates with a protein content of at least 1.9 g protein/100 kcal for the particular nutritional requirements of infants and young children. If the formula under evaluation is considered to be suitable for the particular nutritional requirements of infants and young children, the European Food Safety Authority is asked to advise whether a level of at least 1.9 g protein/100 kcal would be applicable to all formula based on whey protein partial hydrolysates. If this is not the case, the Authority is asked to advise on the special criteria that need to be satisfied for the suitability of such formulae to be demonstrated. ASSESSMENT The assessment of the effects of the addition of Bifidobacterium lactis strain Bb12 to the study formula is outside the terms of reference of the Panel. 1. Introduction The European Union legislation specifies in Directive 91/321/EEC on infant formulae and follow-on formulae that a minimum protein content of 2.25 g/100 kcal is required in infant formulae based on protein partial hydrolysates (protein calculated as total nitrogen times 6.25) and that the formula must provide per energy value at least the same amount of each essential (indispensable) and conditionally essential (indispensable) amino acids as the reference protein in human milk. These values were confirmed in the review undertaken by the Scientific Committee on Food on the Essential Requirements of Infant Formulae and Followon Formulae (SCF, 2003), mainly because at that time no systematic studies to determine the minimal safe protein content of infant formulae with partially hydrolysed protein were available. Page 2 of 16

3 2. Description of the study formula Except for the protein content, the composition of the study formula is in conformity with Directive 91/321/EEC. It has an energy value of 68 kcal per 100 ml of which 46.4% is provided by carbohydrates (lactose and maltodextrin), 45.9% by fat (palm oil, soy bean oil, coconut oil, high oleic acid safflower oil) and 7.6% by protein. The protein content is declared to be 1.9 g/100 kcal or 1.29 g/100 ml, based on a calculation of total nitrogen x If the conversion factor of 6.25 is applied the protein content would amount to 1.86 g/100 kcal, which is closer to the protein content estimated by adding the individual amino acids corrected for the water released in hydrolysing peptide bonds (1.84 g/100 kcal). The protein is a mixture of 63% of a whey protein isolate and 37% of a sweet whey protein concentrate hydrolysed by trypsin. The degree of hydrolysis of the protein source (distribution of length of peptides) is not stated, but the peptide size distribution has been published: 62.8% <1000 Da, 28.6 % Da, 6.9 % Da and 1.8 % >5000 Da (Ziegler et al., 2003). According to the applicant s statement, both protein sources are derived from sweet whey obtained from cows milk after enzymatic (chymosin) precipitation of the caseins. Enzymatic treatment of kappa-casein splits it into precipitating para-kappa-casein and a soluble glycomacropeptide (CGMP) consisting of 63 amino acids, of which eleven are threonine and none is tryptophan (Kielwein, 1994). Sweet whey is then differentially demineralised to obtain CGMP-free whey protein isolate, low in threonine and lactose, and CGMP-containing sweet whey protein concentrate. The relative contribution of these two whey protein products in the formula was chosen to approximate the amino acid composition of human milk protein. Free arginine, histidine and tyrosine have been added. The amino acid pattern of the protein mixture in mg/g crude protein is in conformity with Annex VI of Directive 91/321/EEC and the formula fulfils the requirements of Annex V of the named Directive in providing all amino acids per energy value in at least the amounts of the reference protein human milk. The formula protein does not fulfil the requirements proposed by the SCF (2003) with respect to phenylalanine and tyrosine both per g of protein and per 100 kcal. Whereas in the SCF report a phenylalanine content of 46 mg/g protein and of 83 mg/100 kcal and a tyrosine content of 42 mg/g protein and of 76 mg/100 kcal is proposed, the formula provides of phenylalanine 37 mg/g protein and 68 mg/100 kcal and of tyrosine 34 mg/g protein and 62 mg/100 kcal. The study formula was tested with and without the addition of 2 x 10 7 colony forming units (CFU) of Bifidobacterium lactis strain Bb12 per 1 g powdered formula (2.7 x 10 8 CFU/100 ml formula ready for consumption). A reference formula (Nestlé-Good Start ) based on a protein hydrolysate and with a protein content of 2.44 g/100 kcal is used in a control group in one of two studies reported. According to Ziegler et al. (2003) the protein source in the reference formula was derived from acid whey. It contained less small peptides (<1000 Da) and more large peptides (>5000 Da) than the study formula. 3. Clinical studies with a low-protein infant formula based on partially hydrolysed whey protein Two clinical studies were performed at the Fomon Infant Nutrition Unit at the University of Iowa, USA: Study on food intake and growth in infants fed formulae with reduced protein Page 3 of 16

4 content and with or without added probiotics ; and Gas production and metabolic balance study of infant formulae with or without added probiotics. Both studies are registered in the database Clinitrial. 3.1 Study on food intake and growth in infants fed formulae with reduced protein content and with or without added probiotics This study has been published by Ziegler et al. (2003). The publication contains some information that was not provided by the applicant. Study Objectives - The primary objective was the investigation of growth (weight, length, head circumference) of infants receiving formulae with a protein content of 1.9 g/100 kcal with and without added bifidobacteria in comparison to a reference formula with a standard protein content of 2.44 g/100 kcal. The secondary objectives were the evaluation of formula intake, of the tolerance for the formula, of plasma markers for the dietary nitrogen intake (albumin, total protein, urea nitrogen, haemoglobin and free amino acids), and of the fatty acid composition in phospholipids of red blood cells. Study description - The protocol of a prospective randomised controlled double-blind study was approved by the University of Iowa Committee on Research Involving Human Subjects. Informed written consent by the caregivers of the infants was given. Between January 2000 and May 2001, healthy exclusively formula-fed term infants 6 to 9 days of age were enrolled and randomised via a computer-generated table in the form of a block randomisation to receive one of three formulae to be fed exclusively and ad libitum until the age of 112 days. Study formula 1 (SF1) is described in Section 2. Study formula 2 (SF2) had an identical composition except for the addition of B. lactis Bb12. The reference formula (RF) described in Section 2 had a higher protein content of 2.44 g/100 kcal. There were some minor compositional differences between SF1 and SF2 compared with the RF with respect to some macro- and micronutrients. The formulae were colour coded and the code was not broken before the data had been analysed. Inclusion criteria were healthiness, age <10 days, birth weight 2500 g and 4500 g, gestational age 37 weeks, choice of formula feeding by mother since birth. Rehospitalisation within 10 days after birth and participation in another clinical trial were exclusion criteria. Data collected at baseline included gestational age, weight, length and head circumference at birth, mother s birth date and years of schooling, number of siblings, smoking by the mother or at home, infant s nutrition before trial start, and weight, length and head circumference. Anthropometric data were taken at follow-up visits at age 14 (± 2), 28 (± 2), 42 (± 2), 56 (± 4), 84 (± 4) and 112 (± 4) days, what type of feeding, number of episodes and days with diarrhoea (defined as six or more soft, liquid stools in one day or a gradual or sudden increase in the usual number of stools and/or change in consistency), with cough, fever and hospitalisation and tolerance of feeding were recorded also. The tolerance records over two days included: frequency of stools, colour, consistency and odour of stools and behavioural observations (gas/flatulence, crying, fussing, colic, spitting up, and vomiting). In addition, Page 4 of 16

5 two telephone interviews with the mothers were performed at age 70 ± 4 and 98 ± 4 days (diet compliance, gastrointestinal symptoms, concomitant diseases, medication, and health concerns). Blood samples were taken on days 28, 56, 84 and 112 and analysed for albumin, total protein, urea nitrogen, free amino acids (only days 28 and 112) and fatty acids in erythrocyte phospholipids (only day 112). Formula consumption was calculated from the difference in weight of the formula dispensed and of unused formula returned at each visit. Sample size determination was based on a one-sided testing for differences in weight gain (relevant difference 3 g/day) and assuming a standard deviation of 4.5 g/day for the period of 8 to 112 days of age with a power of 80% at a significance level of 0.05 (AAP, 1988). A gender specific sample size of 28 per group as proposed by AAP was not considered necessary by the investigators because gender effects were taken into account by entering gender in an analysis of covariance (ANCOVA). Statistical analysis - Both intention-to-treat and per-protocol analysis were performed. The latter excluded all subjects which did not complete the trial and those who received additional food. Anthropometric data, average weight and length gain per day and converted into group and gender specific growth curves by linear interpolation for the period of 8 to 112 days were compared between the study groups and to published reference data using ANOVA adjusted for gender. Z-scores were calculated using 2000 Centre for Disease Control and Prevention (CDC) growth charts, and Z-score gains for all available data were compared between the three groups using ANOVA. An additional analysis of growth parameters was performed by fitting a growth curve model to individual serial data using root mean squares error statistic (RMSE), where the RMSE is defined as the square root of the sum of the square differences between the observed and the predicted values from the fitted models. The smaller the RMSE value, the better the fit. Increments in weight or length for each infant at each targeted age were calculated from the difference between the estimated value at the end and at the beginning of the selected age interval and compared using ANOVA. In addition, a random effects model was applied to serial measurements of weight and length measurements for a comparison of formula groups and gender with birth status values included as co-variate. Study population - In total 122 infants (53 boys, 69 girls) with a gestational age of 39.3 ± 1.3 weeks, an average birth weight of 3418 ± 452 g and an average birth length of 50.4 ± 2.4 cm were enrolled. There were six pairs of twins, which were treated as independent infants. Three infants were falsely enrolled (lower gestational age, higher birth weight) and excluded from the per-protocol analysis. Forty infants each received SF1 and SF2 and 42 RF. Gender distribution per formula group was inhomogeneous (females 68% in SF1, 58% in SF2 and 45% in RF at baseline) and was unchanged by dropping out. Anthropometric measurements at baseline were comparable in the three formula groups. Page 5 of 16

6 Twenty-nine infants dropped out of the study, 21 of them before reaching an age of four weeks: eleven in the SF1, nine in the SF2 and nine in the RF group. Reasons for dropping out in the SF1 and RF group were mostly missing of first or other appointment(s), loss of contact, parental decision, non-compliance with exclusive study formula feeding, while in the SF2 group six of nine drop-outs were related to constipation. Three infants inadvertently received incorrect scoops for measuring formula powder resulting in formula of half of the intended concentration. Their formula powder consumption per day, however, did not differ from the average. These infants were excluded from the per-protocol analysis which concerned 88 infants (SF1:27; SF2:28; RF:33). Results - The Panel was not requested to evaluate the addition of bifidobacteria to SF2 but cannot exclude the possibility that differences in outcome between infants receiving SF1 and SF2 can be attributed to this addition. Formula consumption was calculated in g powder/day over the whole study period (103 to 106 days) and corresponded to a formula volume of 910 to 957 ml/day. On average, infants with SF1 consumed 6 g/day (46 ml/day) more formula than infants receiving SF2 and RF. Per-protocol analysis showed that the average weight gain per g formula consumed was somewhat less in the SF1 and SF2 groups (by 8 and 12% in girls and by 7 and 4% in boys, respectively) than in the RF group without reaching statistical significance. Weight - The fitted growth curves after correction for target measurement ages showed (both per-protocol and intention-to-treat analysis) that girls had lower curves for both weight and length than boys, the difference increasing with age. The weight curves of boys receiving different formulae and of girls receiving SF1 and RF did not differ, while girls receiving SF2 showed a flattened growth curve after 56 days. Pair-wise comparison by ANOVA of the weight status values at each of the target ages estimated from the fitted models showed that there was no difference among the formula groups within each gender at each target age. In Table 1, the average weight gain per day calculated from the between-measurement differences are given for the per-protocol analysis. Average weight gain both in per-protocol and intention-to-treat analysis of girls fed SF2 formula is lower than for RF and SF1. Statistically, there was no significant formula effect in both per-protocol and intention-to-treat analysis. Table 1. Average weight gain (g/day), per-protocol analysis Formula Females Males n mean S.D. 95% CI n mean S.D. 95% CI SF SF RF One-month weight gain increments in girls receiving SF2 showed a trend for lower weight gain, however statistically there was no significant difference within each gender among different formula groups. Conversion of the one-month increments into daily weight increments and pair-wise comparison showed the differences to be not greater than the expected 3 g/day for both genders and for the formulae SF1 and RF, and boys receiving SF1 had larger weight gains than those on RF. For SF2 compared with RF this was only found in Page 6 of 16

7 boys, while girls had a smaller gain in weight, which was greater than 3 g/day. The standard deviation for daily weight gain was >4.5, so that the difference in weight gain per day was smaller than the standard deviation of the average weight gain. The Z-scores for weight-for-age and weight-for-length over the whole study period were not significantly different between the formula groups (per-protocol and intention-to-treat analysis). Z-scores converted into Z-score gains (difference of the Z-score obtained at the end of the study and the Z-score obtained at enrolment) were mainly positive, which means a change to a higher growth track, and they were not significantly different between the formula groups by ANOVA. Compared with the US CDC growth charts, the mean percentiles for weight-for-age at baseline were within the 34 th to the 43 rd percentiles for boys and within the 31 st and the 41 st percentile for girls. The corresponding mean Z-scores for boys were to and for girls to At age 112 days, the mean percentiles for boys were within the 53 rd and 65 th percentile and for girls within the 52 nd and 59 th percentile. The corresponding Z-scores were 0.04 to 0.48 for boys and 0.07 to 0.35 for girls. There were no statistically significant differences related to the type of formula. A similar development was observed for the mean percentiles and Z-scores for weight-forlength for boys and girls in all three formula groups, that is somewhat low values relative to US infants at baseline and values within the normal range of growth at the age of 112 days. Random effects model analysis applied to all serial weight measurements with birth weight included as co-variate for determining formula group differences showed that birth weight affected subsequent weight gain and that for boys there was no difference between formula groups. There also was no difference between girls receiving SF1 or RF, while girls receiving SF2 showed slower weight gain compared with RF. Length - Length-for-age curves were almost linear and virtually identical for both females and males in the three groups. Length gain per day was about 1 mm in all groups, somewhat higher in boys than in girls. There was no significant difference between the three formulae even when expressed as length increment per month. Compared with the US CDC growth charts, the mean percentiles at baseline were within the 46 th to the 48 th percentile, and the mean Z-scores within to 0.06 for boys. For girls the corresponding values were within the 35 th to 48 th percentile and the mean Z-scores to At the age of 112 days the mean percentiles for boys were within the 49 th to 61 st percentile and the mean Z-scores 0.0 to For girls the mean percentiles were within the 45 th to the 56 th percentile and the mean Z-scores were to There were no statistically significant differences in the mean percentiles and Z-scores in the three formula groups, both at baseline and at age 112 days using ANOVA followed by pair-wise comparisons. Head circumference - Head circumferences showed no formula-related differences both at baseline and at age 112 days. The Panel notes that the growth patterns of the study infants is comparable to that of US infants as represented in CDC growth charts and that analysis of variance on weight and length at each target age followed by pair-wise comparisons revealed no statistically Page 7 of 16

8 significant differences among the formula groups. The somewhat lower daily weight increment observed only in girls fed SF2, with a difference of >3 g/day compared with RF infants was accompanied by a standard deviation of >4.5 and, therefore presumably not clinically significant. A greater sample size would have been desirable. Protein status - Both for albumin and total protein no effect of the formulae could be demonstrated at all ages, whereas there was a significant increase with age (p <0.0001). Urea nitrogen levels were statistically significantly higher with RF formula (p <0.0001) after logtransformation than with SF1 or SF2. Mean values of urea nitrogen in the SF1 group at the ages of 28, 56, 84 and 112 days of age were 10.34, 9.48, 9.39 and 9.50 mg/100 ml, respectively, somewhat above the levels in breast-fed infants at the same age (7.9, 6.9, 6.8 and 7.0 mg/100 ml) (Fomon, 1993). Type of formula had no significant effect on haemoglobin levels. Haemoglobin increased significantly with age (p <0.0001) after exclusion of the values at 28 days because of high standard deviations. Plasma amino acids - Blood samples were not drawn at defined times after the last feed (2.1 ± 1.2 hours), which complicates the interpretation. Branched-chain amino acids, threonine and phenylalanine were higher in all three groups than in breast-fed infants, and threonine was higher in the RF group than in the infants fed study formula. Tyrosine levels were similar to breast-fed infants. Overall, no excessively low or high amino acid levels compared with breast-fed infants were observed (Bachmann and Haschke-Becher, 2002). Stool characteristics, flatulence and behaviour - The proportion of soft/liquid and foul smelling stools was higher in infants receiving the reference formula. Reports on flatulence per day did not differ in the three groups. In an ANOVA model after square-root transformation, no significant differences between the three formula groups could be found with regard to crying, fussing and spitting up. The incidence of vomiting (0.08 to 0.14 times/day) and of colics (0.3 to 0.34 times/day) was very low and no effect of either formula could be demonstrated. Adverse effects - Nine serious adverse events were reported of which three concerned the incorrect preparation of the formula due to a wrong scoop. The other six events were related to upper respiratory tract infections and/or fever and were judged to be not related to the formula. One hundred sixteen minor adverse events (diarrhoea, flatulence, thrush, constipation, vomiting and gastric reflux) were reported with a non-significant higher incidence in infants receiving RF formula. In a Poisson regression, the number of diarrhoea episodes between two check-ups was significantly higher in infants receiving RF formula (p=0.001). Comments - The study is in conformity with the recommendations of the SCF (2003) on the evaluation of infant formula based on hydrolysed proteins. The study objective of comparing growth rates of infants consuming a formula with a protein level of 2.44 g/100 kcal with ad libitum feeding of a formula based on hydrolysed whey protein at a level of 1.9 g/100 kcal (1.29 g/100 ml) in healthy term infants between the age of eight and 112 days is considered to be obtained. A comparison of growth parameters corrected for gender using ANOVA showed no significant differences between infants consuming study formulae and those consuming reference formula. The Panel is not in a Page 8 of 16

9 position to comment on the trend for slower weight gain in the girls receiving SF2 and the tendency for early drop-outs in this group associated with constipation. There is a trend in all three formula groups for track changes in longitudinal growth from percentiles below 50 for weight-for-age, weight-for-length and length-for-age at baseline to percentiles around or above 50. With regard to the question if these study formulae provide sufficient protein for growth and protein metabolism, the numbers of the two groups consuming study formula can be added together. Blood parameters of protein metabolism were not significantly influenced by feeding formula with a low protein content compared with formula with a standard higher content, except for changes in plasma urea nitrogen, which was closer to values observed in breast-fed infants. The study formulae were well tolerated. Stool consistency was significantly firmer in infants receiving study formula than in infants on reference formula, in whom diarrhoea occurred significantly more frequently. Under the conditions of use, the study formula with reduced protein content is considered safe and promotes normal growth when fed ad libitum. The results from this study, however, do not permit to conclude that all formulae based on hydrolysed whey protein with a content of 1.9 g/100 kcal (or 1.86 g/100 kcal when using a nitrogen conversion factor of 6.25) to be safe and suitable for the particular nutritional requirements of infants. The safety and suitability with respect to protein is a result of both the amount and the quality of the protein. Quality is determined by digestibility and content and pattern of indispensable amino acids. The study formula is based on an enzymatically hydrolysed whey protein mixture with a low content of the threonine-rich caseinoglycomacropeptide and a high content of tryptophan-rich α-lactalbumin and consists predominantly (90%) of peptides with less than 2500 Dalton (Ziegler et al., 2003). The nitrogen content of the study formula is not stated but can be calculated using a conversion factor of 6.38 to be about 203 mg/100 ml, which is quite similar to the nitrogen content of mature human milk of about 180 mg/100 ml (Räihä, 1994) (range , Lönnerdal, 1994). An infant formula with such a low protein/nitrogen content will have to be high in indispensable amino acids and easily digestible/utilisable to provide sufficient nitrogen and indispensable amino acids. The five principal possibilities to determine the protein requirements of infants were outlined by Dewey et al. (1996). 1. A factorial approach which adds up estimates of nitrogen requirements for maintenance and for growth, corrected for conversion efficiency into body proteins. Adequate levels of nitrogen intake were calculated to be 431 mg/kg body weight/day for 0 to 1 month; 326 mg/kg body weight/day for 1 to 2 months; 245 mg/kg body weight/day for 2 to 3 months; 219 mg/kg body weight/day for 3 to 4 months; 200 mg/kg body weight/day for 4 to 5 months and 190 mg/kg body weight/day for 5 to 6 months (Dewey et al., 1996). In order to reach this adequate nitrogen intake during the first month of life more than 200 ml of study formula/kg body weight/day will have to be consumed. Page 9 of 16

10 2. Prospective clinical testing of an infant formula with a defined protein content and quality according to recognised guidelines and under strictly controlled conditions, which has been done with the formula to be evaluated. 3. The model of the exclusively breast-fed infant, assuming that such infants thrive and, therefore can be considered to consume sufficient nitrogen and amino acids. Breastfed infants from 1 to 6 months of age consumed the following amounts of nitrogen: age one month: 362 mg/kg body weight/day; two months: 264 mg/kg body weight/day; three months mg/kg body weight/day; four months: 208 mg/kg body weight/day; six months: 197 mg/kg body weight/day. 4. An approach based on protein-energy ratios (energy percent) when the safe level of protein intake and of the mean requirement of energy intake are not known. Breast-fed infants between the ages of 3 to 4 months achieve crude protein-energy ratios of 8 to 8.5% (Dewey et al., 1996). The study formula under evaluation has a protein-energy ratio of about 7.6%. 5. The metabolic need for individual amino acids can in principle serve to determine the protein requirement. The amino acid amounts needed for body maintenance are, however, not known. By default, the amino acid pattern of human milk protein in combination with observed intakes of human milk or formula by infants growing satisfactorily have been used to calculate requirements for individual amino aids for defined age groups of infants (Dewey et al., 1996; FAO/WHO, 1985; Fomon et al., 1973). When using these data the study formula provides ample amounts of all indispensable amino acids. Infant formulae with intact protein levels at or below 1.9 g/100 kcal were tested in clinical studies by several authors. Fomon et al. (1995) concluded from studies in male infants with formulae with protein density decreasing from 1.63 g/100 kcal at ages 8 to 27 days to 1.31 g/100 kcal at age 84 to 112 days, that these levels were not safe because length gain was reduced compared with infants receiving formula with 2.07 to 2.8 g protein/100 kcal. (The protein levels in this publication were recalculated with the nitrogen conversion factor 6.38 for intact proteins). Räihä et al. (1986) compared growth and blood parameters in infants (ten per group) fed human milk, formula with 1.8 g protein/100 kcal and formula with 2.2 g/100 kcal for three months and found no differences in weight, length, head circumference and serum albumin. During the first month serum urea nitrogen and urinary nitrogen excretion were lower in infants fed the low-protein formula than in breast-fed infants. Räihä (1994) therefore recommended graded protein levels in infant formula during the first half year of life: first month 2.2 g protein/100 kcal; second month 2 g protein/100 kcal; third month 1.8 g protein/100 kcal; fourth month 1.6 g protein/100 kcal. A cows milk-based formula with a protein content of 1.8 g/100 kcal (recalculated) was fed to 16 infants between 8 and 112 days of age in comparison with breast-fed and standard formula-fed infants. It was considered not to be safe because infants increased their energy intake significantly when consuming this formula (Fomon et al., 1999). Page 10 of 16

11 When infants (29 per group) were fed one of three formulae (two with 1.9 g protein/100 kcal; one with 2.3 g protein/100 kcal) from birth to 4 months there were no differences in nitrogen retention, weight and length increment and serum albumin (Räihä et al., 2002). In another study, two whey protein dominant formulae with protein contents of 1.83 g/100 kcal and 2.24 g/100 kcal were fed to normal infants between the ages 28 to 128 days in a cross-over design and nitrogen retention was found to be similar (Ziegler, 2002). In conclusion, some recent studies with infant formulae with a protein content of 1.8 to 1.9 g/100 kcal of (modified) intact cows milk protein have been shown to be adequate to promote normal growth. On the other hand, formulae based on hydrolysed proteins, which have been clinically tested in the past, contained regularly more protein than those based on intact protein because of uncertainty about their biological equivalence due to reports on reduced growth and nitrogen utilisation with some such formulae (Rigo et al., 1995). 3.2 Gas production and metabolic balance study of infants fed formulae with or without added probiotics The study was carried out in healthy term infants enrolled between May 2000 and September 2001 and approved by an ethics committee. Informed written consent was obtained from the parents/care givers. Objectives - The primary objective was a comparison of the nitrogen balance in infants fed sequentially two formulae with reduced protein content (1.9 g/100 kcal) with or without added Bifidobacterium lactis strain Bb12. The study formulae were the same as in the previous study. Secondary objectives were the comparison of the absorption and retention of calcium, magnesium, phosphorus and zinc and of the absorption of fat, of the gas production (CO 2, H 2, CH 4, H 2 S and methanethiol MES) from faeces, of the microbial analysis of stools, of stool characteristics and of feeding-related behaviour in infants receiving the two study formulae. For stool parameters a comparison with breastfed infants was also performed. The results of these investigations are not relevant for the task of the Panel. Study design - The study was double-blind, randomised, controlled and performed as a cross over trial with colour coded SF1 and SF2 sequentially, exclusively and ad libitum. After at least ten days of formula feeding stools were obtained and balance studies which lasted three days were performed. In addition, healthy exclusively breast-fed infants (less than two formula feedings per week) were enrolled and provided stool samples only. Data collected at baseline were gestational age, type of delivery, date of birth, sex, birth weight and length, APGAR scores, number of siblings, history of colic in sibling, history of feeding. Data collected during three days before balance studies or stool collection related to feeding history, number of feedings of formula or breast-milk, volume of formula per meal, intake of other types of food or liquids, stool characteristics and feeding-related behaviour. During the balance study body weight and formula intake were recorded. Urine and faeces were collected. Absorption (intake minus faecal excretion) and retention (intake minus Page 11 of 16

12 excretion via urine and faeces) were calculated and expressed as a percentage of intake and in absolute amounts per body weight. Statistics - Sample size determinations to achieve a power of 80% and a probability of error of 5% in one-tailed testing were based on an assumed relevant difference in nitrogen retention of 40 mg/kg body weight/day and a standard deviation of 32 mg/kg body weight/day and required 8 infants for the balance studies. Data were analysed by intention-to-treat and perprotocol, excluding infants with premature completion, antibiotic therapy and consumption of additional formula or food. Data from the infants consuming SF1 and SF2 were analysed by repeated measures ANOVA (within infants) with a test for carry-over (between groups). Study population - Nineteen healthy exclusively formula-fed term infants were enrolled at a mean age of 94 ± 42.9 (range 21 to 111) days and were randomised to one of two possible sequences of formula SF1 and SF2. There were three drop-outs, two because of parents decision, one because of antibiotic therapy. Two sequential balance studies were performed in eight infants. Eleven breast-fed infants were enrolled, of which two stopped before the first stool collection because of parents decision and because of non-compliance with the study protocol. Twenty-five infants (n=16 formula; n=9 breast milk) completed the study and could be analysed per-protocol. Gestational age was not different between formula- and breast-fed infants and the groups were comparable with respect to body weight, length and BMI at birth. Compliance was excellent. Results Balances - The data on absorption and retention of nitrogen per formula group are presented in Table 2. Table 2. Nitrogen balances in eight infants with study formula 1 and 2 (mean ± SD) Nitrogen balance SF1 SF2 Difference 1-2 P-value (mg N/kg body weight/day) Absorption ± ± ± Retention ± ± ± (% of nitrogen intake) Absorption 85.5 ± ± ± Retention 37.9 ± ± ± There were no statistically significant differences in nitrogen absorption and retention and absorption and retention rates between the period when infants consumed SF1 compared with the period when they consumed SF2. Absorption of nitrogen was about 300 mg/kg body weight/day (86% of intake) and retention was about 141 mg/kg body weight/day (38% of intake) and decreased with age (from more than 40% at the age of one month to 25% at the age of 5 months. The Panel calculated the individual data on volume and nitrogen intake from the data provided by the applicant and related them to the individual body weight and age. Page 12 of 16

13 Individual levels of nitrogen intake differed widely between 584 (in a one-month old infant) and 240 mg/kg body weight/day (in a five-month old infant), which would correspond to a crude protein (conversion factor 6.25) intake of 3.7 g and 1.5 g/kg body weight/day and a formula volume of 288 and 118 ml/kg body weight/day, respectively. All infants between four and seven weeks of age consumed more than 200 ml of formula/kg body weight/day, while infants between 18 and 21 weeks of age consumed between 120 to 180 ml/kg body weight/day. From the individual data on nitrogen retention and absorption, nitrogen utilisation (retention divided by absorption x 100) can be calculated. It ranges from 29 to 67% (mean value 45% both for SF1 and SF2) and is higher in younger infants than in older infants. This is within the range of values reported for conventional formula during the first month of life (66%) and lower than reported for human milk (77%) (Rigo et al., 1995). Stool frequency - Stool frequency was significantly higher in breast-fed than in SF1-fed infants (3 ± 1.25 versus 1.71 ± 0.47/day). Safety and tolerance - There were no significant differences between the formula groups. Breast-fed infants had more frequent stools and all of them were of soft consistency and smelled normal. Flatulence episodes were more frequent in infants consuming formulae than in breast-fed infants. There were no significant differences in spitting up, colic and vomiting between infants consuming formula compared with infants consuming breast-milk. Adverse events - No serious adverse events happened during this trial. Comments - The primary objective of this study was to evaluate if the addition of Bifidobacterium lactis strain Bb12 to the study formula had an influence on the balance of nitrogen, calcium, magnesium, phosphorus and zinc and on the absorption of fat. This information is of very restricted usefulness for the task of the Panel. There were no statistically significant differences in nitrogen absorption and retention when infants consumed SF1 and SF2, therefore the Panel considers that the outcomes of the two different study formulae may be considered together. Nitrogen balance data showed a high absorption of dietary nitrogen (86% of intake) and an average retention of 38% of the intake. The intake of nitrogen was highly variable and the formula volume necessary to obtain the highest reported nitrogen intake (288 ml/kg body weight/day) in a one-month old infant is high. Nitrogen utilisation varied between 29 and 67% (mean 45%) and is within the range reported in full-term infants fed conventional formula during the first month of life (66%). It is lower than in breast-fed infants (77%). Earlier reports on reduced nitrogen utilisation in formula based on hydrolysed protein concerned feeding on whey-hydrolysate formula (1.7 g protein/100 ml) and this was accompanied by reduced gain in weight, length and head circumference (Rigo et al., 1994). Stools from formula-fed infants were less frequent, tended to be firmer and more often smelled foul than those from breast-fed infants. Page 13 of 16

14 CONCLUSIONS AND RECOMMENDATIONS The conclusions of the Panel are as follows: - When fed ad libitum, the formula is as suitable to satisfy the particular nutritional requirements of young infants and as safe as a formula based on hydrolysed whey protein with a higher protein content. This conclusion is primarily based upon the growth data of term-infants who were studied between the ages of 8 to 112 days. The nitrogen balance data and the protein status parameters are considered to be supportive of this conclusion. The conclusion on the suitability of the formula with a crude protein content of 1.9 g/100 kcal is specific for this formulation and the protein source it contains. While no data on follow-on formula based on hydrolysed whey protein with a protein content of 1.9 g/100 kcal have been submitted, the Panel considers that a formula with this protein formulation is suitable for use in older infants in conjunction with complementary foods. - Any infant formula with a similar crude protein content but differing in protein quality, protein sources or protein processing needs to be tested in a clinical study and its safety should be documented and assessed. - Recommendations on how to assess the protein quality, sufficiency and suitability in infant formulae have been published. They include a clinical trial of three to four months duration in a group of healthy term infants with a number to provide sufficient statistical power in comparison to an existing infant formula and/or a reference group of breast-fed infants, and starting preferably at or near birth. Data on growth, protein status and on formula intake should be collected (SCF, 2003; Aggett et al., 2001 and 2003; AAP, 1988). DOCUMENTATION PROVIDED TO EFSA Dossier submitted by Nestlé to the European Commission in support of the safety and nutritional efficacy of a new low-protein infant formula based on whey protein partial hydrolysates, with or without added Bifidus cultures. September Additional information submitted by Nestlé. Addendum. August AAP (American Academy of Pediatrics) (1988). Committee on Nutrition. Clinical testing of infant formulas with respect to nutritional suitability for term infants. Prepared under FDA contract Elk Grove Village, IL, USA. REFERENCES Aggett PJ, Agostoni C, Goulet O, Hernell O, Koletzko B, Lafeber HN, Michaelsen KF, Rigo J, Weaver LT (2001). The nutritional and safety assessment of breast milk substitutes and other dietary products for infants: A commentary by the ESPGHAN Committee on Nutrition. J Pediatr Gastroenterol Nutr 32: Page 14 of 16

15 Aggett PJ, Agostoni C, Axelsson I, Goulet O, Hernell O, Koletzko B, Lafeber HN, Michaelsen KF, Morley R, Rigo J, Szajewska H, Weaver LT (2003). Core data for nutrition trials in infants: a discussion document - A commentary by the ESPGHAN Committee on Nutrition. J Peditar Gastroenterol Nutr 36: 1-5. Bachmann C and Haschke-Becher E (2002). Plasma amino acid concentrations in breast-fed and formula-fed infants and reference intervals. In: Räihä NCR and Rubaltelli FF (Eds) Infant formulae: closer to the reference. Nestlé Nutrition Workshop Series Vol 47, Williams & Wilkins, Philadelphia. Commission Directive 91/321/EEC on infant formulae and follow-on formulae. Official Journal of the European Communities, , L 175, p 35. Commission Directive 96/4/EC amending Directive 91/321/EEC on infant formulae and follow-on formulae. Official Journal of the European Communities, , L 49. Dewey KG (2001). Nutrition, growth, and complementary feeding of the breastfed infant. Pediatr Clin N America 48 (Part I): Dewey KG, Beaton G, Field C, Lönnerdal B, Reeds P (1996). Protein requirements of infants and children. Eur J Clin Nutr 50: S119-S150. FAO/WHO (Food and Agriculture Organisation/World Health Organization) (1985). Joint FAO/WHO/UNU Expert Consultation. Energy and Protein Requirements. WHO Technical Report Series, No Fomon SJ (1993). Nutrition of normal infants. St. Louis, Mosby. Fomon SJ, Thomas LN, Filer LJ, Anderson TA, Bergmann KE (1973). Requirements for protein and essential amino acids in early infancy. Acta Paediatr Scand 62: Fomon SJ, Ziegler EE, Nelson SE, Frantz JA (1995). What is the safe protein-energy ratio for infant formulas? Am J Clin Nutr 62: Fomon SJ, Ziegler EE, Nelson SE, Rogers RR, Frantz JA (1999). Infant formula with proteinenergy ratio of 1.7 g/100 kcal is adequate but may not be safe. J Pediatr Gastroenterol Nutr 28: Harzer G (1989). Über die Zusammensetzung von Muttermilch - Zur Adaptation von Säuglingsnahrungen. In: Milchwissenschaften, Vol. 7. E Renner (Ed.) Gießen. Kielwein G (1994). Leitfaden der Milchkunde und Milchhygiene. 3rd Edition. Blackwell Wissenschaftsverlag, Berlin. Lönnerdal B (1994). Nutritional importance of non-protein nitrogen. In: Protein Metabolism During Infancy. Räihä NCR (Ed.) Nestlé Nutrition Workshop Series, Vol. 33, Raven Press, New York, pp Page 15 of 16

16 Räihä NCR (1994). Protein content of human milk, from colostrum to mature milk. In: Protein metabolism during infancy. Räihä NCR (Ed). Nestlé Nutrition Workshop Series, Vol. 33, Raven Press, New York, pp Räihä NCR, Fazzolari-Nesci A, Cajozzo C, Puccio G, Monestier A, Moro G, Minoli I, Haschke-Becher E, Bachmann C, Van t Hof M, Carrié Fässler A-L, Haschke F (2002). Whey predominant, whey modified infant formula with protein/energy ratio of 1.8 g/100 kcal: adequate and safe for term infants from birth to four months. J Pediatr Gastroenterol Nutr 35: Räihäe N, Minoli I, Moro G (1986). Milk protein intake in the term infant. I. Metabolic responses and effects on growth. Acta Paediatr Scand 75: Rigo J, Salle BL, Picaud JC, Putet G, Senterre J (1995). Nutritional evaluation of protein hydrolysate formula. Eur J Clin Nutr 49: S26-S38. Rigo J, Salle BL, Putet G, Senterre J (1994). Nutritional evaluation of various protein hydrolysate formulae in term infants during the first month of life. Acta Paediatr 402: Rudloff S and Kunz C (1997). Protein and non-protein nitrogen components in human milk, bovine milk and infant formula: quantitative and qualitative aspects of infant nutrition. J Pediatr Gastroenterol Nutr 24: SCF (Scientific Committee on Food) (2003). Report of the Scientific Committee on Food on the revision of essential requirements of infant formulae and follow-on formulae (adopted on 4 April 2003). Ziegler EE (2002). Protein requirements in infancy. In: Infant formula: closer to the reference. Räihä NCR and Rubaltelli FF (Eds). Nestlé Nutrition Workshop Series, Vol. 47 Supplement. Williams & Wilkins, Philadelphia, pp Ziegler EE, Jeter JM, Drulis JM, Nelson SE, Haschke F, Steenhout P, Brown C, Maire J-C, Hager C (2003). Formula with reduced content of improved, partially hydrolyzed protein and probiotics: infant growth and health. Monatsschr Kinderheilkd 151: S65-S71. PANEL MEMBERS Wulf Becker, Francesco Branca, Daniel Brasseur, Jean-Louis Bresson, Albert Flynn, Alan A. Jackson, Pagona Lagiou, Martinus Løvik, Geltrude Mingrone, Bevan Moseley, Andreu Palou, Hildegard Przyrembel, Seppo Salminen, Stephan Strobel, Henk van den Berg, and Hendrik van Loveren. Page 16 of 16