protein concentration in milk formula, growth, and …jn.nutrition.org/content/146/3/551.full.pdfthe...

The Journal of Nutrition

Nutrition and Disease

Protein Concentration in Milk Formula, Growth,and Later Risk of Obesity: A SystematicReview1–3

Bernadeta Patro-Go1ab,4* Bart1omiej M Zalewski,4 Stefanie MP Kouwenhoven,5 Jacek Karas,4

Berthold Koletzko,6 Johannes Bernard van Goudoever,5,7 and Hania Szajewska4

4Department of Pediatrics, The Medical University of Warsaw, Warsaw, Poland; 5Department of Pediatrics, VU University Medical

Center, Amsterdam, Netherlands; 6Ludwig Maximilians University, Dr. von Hauner Children�s Hospital, Division of Metabolic and

Nutritional Medicine, University of Munich Medical Centre, Munich, Germany; and 7Department of Pediatrics, Emma Children�sHospital, Amsterdam Medical Center, Amsterdam, Netherlands

Abstract

Background: Protein intake may influence important health outcomes in later life.

Objective: The objective of this study was to investigate current evidence on the effects of infant formulas and follow-on

formulas with different protein concentrations on infants� and children�s growth, body composition, and later risk of

overweight and obesity.

Methods: In this systematic review, we searched electronic databases (including MEDLINE, Embase, and the Cochrane

Library) up until November 2014 for randomized controlled trials (RCTs). Eligible studies had to include children aged 0–3 y

who represented the general population and were fed cow milk–based infant formulas with variations in protein

concentration. Control groups received lower-protein cow milk–based formulas (as defined by the authors). The primary

outcomeswere growth, overweight, obesity, and adiposity. Various time points for outcomes assessment were accepted

for inclusion. If possible, a meta-analysis was performed.

Results: Twelve RCTsmet our inclusion criteria. Different formula protein concentrations did not affect linear growth other

than a transient effect on mean length at 3 mo observed in a meta-analysis of 4 studies (mean difference, – 0.27 cm; 95%

CI: 20.52, 20.02). Lower mean weight and weight z scores obtained from the infants fed lower-protein formulas were

observed only from 6 to 12mo of age. Data from one large RCT showed that consumption of a lower-protein infant formula

may reduce bodymass index at 12mo of age and later (12mo, 24mo, and 6y) and the risk of obesity at 6 y. Effects on body

composition remained unclear.

Conclusions: The current evidence is insufficient for assessing the effects of reducing the protein concentration in infant

formulas on long-term outcomes, but, if confirmed, this could be a promising intervention for reducing the risk of

overweight and obesity in children. In view of the limited available evidence, more studies replicating effects on long-term

health outcomes are needed. J Nutr 2016;146:551–64.

Keywords: RCT, systematic review, infant nutrition, pediatrics, protein

Introduction

Overweight and obesity are among the most serious publichealth challenges. The effectiveness of different treatment

modalities is limited, and, thus, interest in the identification of

modifiable risk and protective factors is growing. One of these

includes research on the potential role of infant feeding on body

composition and the likelihood of being overweight or obese in

later life.Sufficient protein intake early in life is of major importance.

Deficient protein intake can lead to suboptimal growth and

impaired neurodevelopment (1). However, protein intake that is

too high may have adverse effects as well. It has been

documented that rapid weight gain in infancy is associated

with an increased chance of later obesity (2), and weight gainfrom birth to 24 mo is the best overall predictor of lateroverweight (3). High early protein intake in excess of metabolicrequirements may stimulate the secretion of insulin and insulin-like growth factor I (4), thereby enhancing weight gain ininfancy and increasing the later risk of obesity. This is known asthe early protein hypothesis (5), and it is based on earlyobservations by Rolland-Cachera et al. (6). These investigatorsrelated a high protein intake during early childhood to modifiedendocrine responses and an increased risk of obesity at schoolage. The mechanisms by which increased protein intake affectsweight gain and body composition are not yet completely clear.However, the endocrine and metabolic responses of infants are

ã 2016 American Society for Nutrition.

Manuscript received September 8, 2015. Initial review completed October 20, 2015. Revision accepted January 8, 2016. 551First published online February 10, 2016; doi:10.3945/jn.115.223651.

Downloaded from https://academic.oup.com/jn/article-abstract/146/3/551/4578260by gueston 17 June 2018

affected substantially by dietary protein intake (7). Therefore,lowering protein intake from milk formulas may exhibitbeneficial effects on weight gain, body composition, andmetabolic diseases in later life.

There is currently no consensus regarding what should bethe appropriate amount of protein in infant formulas (8–12).European Directive 2006/141/EC on infant and follow-onformulas (13), as well as the FDA (14), Codex Alimentarius(15), and European Food Safety Authority (16), define theminimum protein-to-energy ratio for infant formulas based oncow milk and soy protein as 1.8 g/100 kcal and 2.25 g/100 kcal,respectively, but they do not define an optimal intake amount.

In 2014, the European Food Safety Authority carried out aliterature search and review as preparatory work for theevaluation of the composition of infant and follow-on formulasand growing-up milk. With regard to reduced-protein formulas,this report concluded that neither negative health effects norclear benefits from the use of this type of formula could beestablished (8). Because there is uncertainty regarding the effectof dietary protein intake in childhood on growth and bodycomposition, and the later risk of overweight, obesity, andmetabolic syndrome, we aimed to investigate systematically thecurrent evidence on this proposed relation. The review protocolwas registered on the PROSPERO International ProspectiveRegister for Systematic Reviews (crd.york.ac.uk/prospero/index.asp) as CRD42014015573.

Methods

Inclusion/exclusion criteria. Study inclusion and exclusion criteria are

summarized in Table 1 (CRD42014015573).

Search methods for identification of studies. The followingelectronic databases were searched until November 2014 for published

studies that fulfilled our criteria: Cochrane Central Register of Controlled

Trials, PubMed (including MEDLINE), Embase, and the CumulativeIndex to Nursing and Allied Health Literature. To identify potential

systematic reviews/meta-analyses, we browsed The Cochrane Database

of Systematic Reviews and Database of Abstracts of Reviews of Effects.

Four reviewers (BP-G, BMZ, SMPK, and JK) were involved in the

process of the search, which was carried out independently by at least 2

reviewers. No language restrictions were applied. A detailed searchstrategy (Supplemental Table 1) was prepared with the support of an

information specialist from the Centre for Reviews and Dissemination,

University of York. Initially, the title, abstract, and keywords of every

record identified with the use of our search strategy were screened.Irrelevant articles were excluded by title or abstract. Full texts were

obtained for all potentially relevant studies. Differences between

reviewers were resolved by discussion until a consensus was reached.

The reference lists from identified studies and key review articles, as wellas selected trial registries (clinicaltrials.gov, www.clinicaltrials.gov; EU

Clinical Trials Register, www.clinicaltrialsregister.eu), were searched.

Proceedings from major scientific gastrointestinal and nutritionalmeetings published in the last 3 y also were screened. Finally, an

attempt was made to obtain additional data by direct contact with

experts in the field.

Data collection and analysis. An initial screening of the title, abstract,

and keywords of every record identified was performed. The next step

was to retrieve the full text of potentially relevant studies. Three

reviewers (BP-G, BMZ, and JK) independently assessed the eligibility ofeach potentially relevant study with the use of the inclusion criteria. All

cases in which different opinions or questions about the eligibility of a

particular study occurred were discussed with at least one other memberof the group (usually the expert of the team).

Data extraction andmanagement.With the use of a previously prepared

template of a table of evidence, 2 reviewers independently extracted thedata necessary for study characteristics and risk of bias assessment.

Assessment of risk of bias. The Cochrane Collaboration�s tool for

assessing risk of bias was used while including the following criteria:adequacy of sequence generation; allocation concealment; blinding of

participants, personnel, and outcome assessors; incomplete outcome

data; and selective reporting.

Measures of treatment effect. If feasible, for the continuous

outcomes, the results have been reported as the mean difference (MD)8

between the lower-protein formula groups and higher-protein formulagroups with a 95% CI. For the dichotomous outcome (prevalence of

obesity), the results have been reported as the RR between the lower-

protein and higher-protein formula groups with a 95% CI. Missing SDs

in 2 randomized controlled trials (RCTs) were calculated based on thereported SEMs with the use of the method described in the Cochrane

Handbook for Systematic Reviews of Interventions (17). For outcomes

expressed with the use of different units, if possible, data were

transformed for their unification (e.g., millimeters into centimeters).Four authors were contacted by e-mail or personally to obtain missing

data or clarify equivocal issues. This resulted in our obtaining additional

unpublished data from 2 published trials (18, 19).

Assessment of heterogeneity. Clinical and methodologic inconsistency

between the selected studies was assessed during the initial analysis, data

extraction, and quality assessment process. The decision to pool datatogether into a meta-analysis was based mainly on this assessment.

Additional information about heterogeneity was obtainedwhen the results

of the chi-square test and I2 value were analyzed for some outcomes.

Assessment of reporting biases. Although originally planned,

because of the insufficient number of studies included in the meta-

analysis (<10), we did not create a funnel plot, nor did we performa statistical test for its asymmetry to detect potential publication bias.

Data synthesis. A narrative synthesis was undertaken if conditions

required for performance of a meta-analysis were not fulfilled. For

1 The research leading to these results received funding from the European

Union’s Seventh Framework Programme (FP7/2007-2013), project EarlyNutrition

under grant agreement no. 289346. The study formula for the Childhood Obesity

Programme study was produced by Bledina, Steenvorde, France as part of a

contract with the European Commission.2 B Patro-Go1ab and BM Zalewski participated in a conference sponsored by

Nestle. SMP Kouwenhoven is a PhD student involved in a study designed to look

at the effects of a low-protein formula on weight gain and body composition. B

Koletzko is a member of the National Breastfeeding Committee and tends to be

biased toward breastfeeding. Both B Koletzko and his employer, the

Ludwig-Maximilians-University of Munich, Germany, have received support for

scientific and educational activities by companies that market products for

infants and children, including Abbott, Baxter, B. Braun, Dairy Goat Cooperative,

Danone, Fresenius Kabi, Fonterra, Hipp, Mead Johnson, Nestle, and Yakult,

predominantly as part of publicly funded research projects with the support of

the European Commission or the German government. JB van Goudoever holds

patents for a specific blend of essential amino acids in order to develop a new

infant formula, and is the sponsor of a study funded by the European Union that is

designed to look at the effects of a low-protein formula on weight gain and body

composition. He participated as a clinical researcher, advisory board member,

and/or speaker for Danone, Dicofarm, Ferrero, Hipp, Nestle Nutrition Institute,

Nutricia, and MJN. H Szajewska has participated as a clinical investigator,

advisory board member, and/or speaker for Arla, Biogaia, Biocodex, Danone,

Dicofarm, Hipp, Nestle, Nestle Nutrition Institute, Nutricia, Mead Johnson,

Merck, Sequoia, and Yakult. J Karas, no conflicts of interest.3 Supplemental Figure 1 and Supplemental Tables 1–3 are available from the

"Online Supporting Material" link in the online posting of the article and from the

same link in the online tables at http://jn.nutrition.org.

*To whom correspondence should be addressed. E-mail: [email protected].

8 Abbreviations used: CHOP, EU Childhood Obesity Programme; MD, mean

difference; RCT, randomized clinical trial.

552 Patro-Go1ab et al.


quantitative analysis and pooling individual study data together, we used

the ReviewManager computer program, version 5.3 (The Nordic Cochrane

Centre; The Cochrane Collaboration). For the 4 multiarm studies included in

the meta-analysis (18, 20–22), in order to avoid a unit-of-analysis error, wecombined the study groups to achieve a single pairwise comparison while

following the formulas provided by the Cochrane Handbook for Systematic

Reviews of Interventions (17). Some studies, although reporting an outcomeof interest, were excluded from the meta-analysis because of a lack of

necessary data (a lack of sample size, narrative results description without

numerical data, etc.). The authors of one trial assessed the outcomes for

lower- and higher-protein formula groups at different ages of children; thus,this trial was excluded from the analysis (23). In the case of another study, we

could not rule out the possibility of the presentation of duplicated data (21).

However, because data from this study were not pooled together in the meta-

analysis with data provided by other authors, the inclusion of this study isunlikely to result in biased estimates. The study aimed to compare infants that

were given different weaning foods (Swedish or Mediterranean); however,

children in both groups were randomly assigned to receive formula with alower or higher protein concentration. Because the authors provided

sufficient data, we compared the growth of infants within each study

group (different protein concentrations), but not between the groups

themselves (different weaning food). In the case of one trial [Raiha et al.(24)], when presenting the results, we made an assumption that the mean

values were accompanied by SD values (but not SEM values). However,

this was not clearly stated in the original paper.

Results

For a PRISMA flow diagram documenting the identificationprocess for eligible trials, see online Supplemental Figure 1. For

the characteristics of the excluded trials, with reasons forexclusion, see online Supplemental Table 2. The characteristicsof 17 included publications describing 12 trials (10 RCTs and 2quasi-randomized trials) (18–34) are summarized in Table 2. Sixof the 17 selected publications reported the results of the EUChildhood Obesity Program (CHOP) study (19, 30–34).

All study participants were healthy, term infants. Themajority of children were from birth to 4 mo of age atenrollment and exclusively formula-fed during the first monthsof life. Except for 2 trials conducted in China (25, 26), theincluded trials were undertaken in Europe.

The protein concentration in different infant formulas variedgreatly between the studies, ranging from1.1 to 2.1 g/100mL in thelower-protein formula group to 1.5 to 3.2 g/100 mL in the higher-protein formula group. In one trial (28), the protein concentrationwas unreported (described as low- and high-protein formulas).Differences in the types of protein used with respect to the whey-to-casein ratio were observed between the studies and between thestudy groupswithin individual studies. Some of the included studieswere multiarm studies with >2 intervention groups. Two separatestudies by Lonnerdal and Chen (25, 26) involved the samepopulation of infants observed during different time periods(from birth and from 4 mo of age). In addition, in many trials, areference group of breastfed infants was recruited. The durationof the intervention usually lasted between 3 and 5 mo, with theexception of that in the CHOP study, in which infants receivedstudy formula until 12 mo of age. Also, The CHOP study was theonly identified study with a longer (>12 mo of age) follow-up.

TABLE 1 Study inclusion and exclusion criteria

Inclusion criteria Exclusion criteria

Types of study Randomized controlled trials and quasi-randomized trials

were included

Types of participants Infants and young children (ages 0–3 y), representing the

general population

Preterm infants

Children with chronic diseases

Types of intervention Cow milk–based infant formulas and/or follow-on formulas

with variations in protein quantity

Duration of intervention ,3 mo

Mixed feeding groups (breastfed and concurrently formula-

fed children), if formula feeding was a predominant

source of milk

Formula with a protein from a different source

(soy, rice, hydrolyzed, elemental, etc.)

The differences in composition of amino acids and/or

macronutrients that were secondary to the protein

amount modification (in order to achieve an isocaloric

formula) were not considered as an actual additional

modification that disqualifies the study from inclusion

Milk formula with additional modifications other than

quantitative protein (i.e., addition into milk formula of

extra components such as a-lactalbumin, lactoferrin,

long-chain PUFAs, prebiotics, probiotics, synbiotics, fat

globule membranes, tryptophan, taurine, or cholesterol).

Types of comparison Lower-protein cow milk–based infant formulas and follow-

on formulas

Types of outcome

Primary outcome measures Obesity and overweight, as defined by the authors

Anthropometric growth markers: BMI, weight, length/

height, head circumference, or changes in these markers

at different time intervals (as expressed by the authors)

Fat mass and fat-free mass (measured with the use of

different methods/devices)

Secondary outcome measures1 Biomarkers of disease (i.e., concentrations of TGs; HDL,

LDL, and total cholesterol; and fasting glucose)

Metabolic syndrome criteria

Concentrations of insulin-like growth factor I and insulin

Other data reported by the investigators if relevant to the

current review

1 Studies with various time points for outcomes assessment (directly after the intervention or after a different follow-up period) were accepted for inclusion.

Protein concentration of infant formulas 553


TABLE

2Characteristicsofincludedstudies1

Reference

Stud

ytype

Samplesize

andsetting

Population

Lower-protein

group

Higher-protein

group

Outcomes

relevant

forthisreview

Effect

size

Funding

Comments

Zoppietal.,

1978

(29)

Quasi-RCT

n=55,Italy

Healthy,term,

nonbreastfe

d

infants

n=28;com

mercia

llyavail-

ablepowderedmilk

(whey:caseinratio

18:82)

fedexclu

sivelyup

to179d

oflife;proteinconcen-

tration15%:2.1gand66

kcal/100

mL(3.2g/100

kcal)

n=27;com

mercially

availablepowdered

milk

(whey:casein

ra-

tio18:82);proteinata

15%concentration:3.0g

and64.2kcal/100

mL

(4.7g/100kcal)

Weight,length,and

their

gainmeasuredevery2

wk(dataprovided

for

days

164,194,and301)

Meanweightandmean

gaininweightslightly

lower

inlower-protein

groupat

ages

164–

301d,with

statistically

significant

diffe

rence

at5.5moof

ageonly;

meangaininlengththe

same(NS)

Not

reporte

dThestudywas

designed

toassess

therelation

betweenbloodg-glob-

ulinconcentra

tions

and

dietaryproteinintake

Age:fro

mday1

oflife

Higher

amount

ofcarbo-

hydrates

inlower-pro-

teingroups

Childrenwith

thalassemia

andillnesses

requiring

hospitalizationnotin-

cluded

Zoppietal.,

1982

(23)

Quasi-RCT

n=41,Italy

Healthy,term

infants

n=7;fedexclusivelywith

cow-milk

formulawith

a

proteincontentof

1.6g/

100mL(2.2g/kcal);74.2

kcal/100

mLfor4.5mo


cow-milk

formulawith

aproteincontentof3.0g/

100mL(4.0g/kcal);

75.4kcal/100

mL

Weightandlength

mea-

suredevery2wk

Grow

th(weight,length,

andhead

circum

fer-

ence)normalandsimi-

larinallinfants**

Research

supportedin

partby

CNRandMPI

grants

Thestudywas

designed

toassess

immunocom

-

petence

Age:fro

mbirth

Serumcholesteroland

TGconcentra

tions

assessed

at4.5mo

ofage

Observed

concentra

tions

reflected

theam

ount

andquality

ofprotein

intake;TGconcentra

-

tions

withinthenormal

range**

Theaimof

thestudywas

tocompare

cow-milk

formulaandsoyfor-

mula;hospitalized

in-

fantswereexcluded

Meandaily

gaininweight

andlength,and

mean

increase

inhead

cir-

cumference

Nodiffe

rencebetween

groups

Zoppietal.,

1983

(28)

RCT

n=62,Italy

Healthy,term

infantswho

wereto

be

artificially

fed

Fedexclusivelywith

low-

proteincowmilk–based

formula(n=9)oradapted

accordingto

ESPGHA

N

recommendations–based

formula(n=10)for

5mo


high-proteincowmilk–

basedformula

Meandaily

gainsin

weight,length,and

head

circum

ference

Meandaily

gainsin

weight,length,and

head

circum

ference:

``normal

``and

``similar

toearlier

findings

``**

Not

reporte

dThestudywas

designed

toassess

theantibody

response

todiffe

rent

vaccines;group

4re-

ceivedsoyformula(n=9);

27infantsw

erebreastfed

Age:fro

mbirth

Theexactamount

ofprotein

intheformulawas

not

reporte

d

Theexactam

ount

ofpro-

teinintheformulawas

notreporte

d

(Continued)



TABLE

2Continued

Reference

Stud

ytype

Samplesize

andsetting

Population

Lower-protein

group

Higher-protein

group

Outcomes

relevant

forthisreview

Effect

size

Funding

Comments

R€ aih€ aet

al.,

1986

(24)

RCT

n=30,Italy

Healthy,term

infants

n=10;formulawith

a

proteincontentof

1.2g/

100mL(whey:caseinra-

tio60:40),67kcal/100

mL,givenfor12

wk

n=10;formulawith

a

proteincontentof

1.6

g/100mL(whey:casein

ratio

60:40),67kcal/

100mL

Meanweight,length,and

head

circum

ferenceat

1,2,4,8,and12

wkof

age;

NS

Studysupportedby

Findus,B

juv,Sw

eden;

Nestle,Vevey,

Switzerland;and

The

NationalR

esearch

Council,Italy

Third

armof

study—

breastfedinfants;20

infantsinformula-fed

groups

(unclear

whether

thenumberof

subjectswas

thesame

ineach

group)

Age:fro

mbirth

Grow

thrate

from2to

12

wkcomparedonlywith

human

milk

group

NS

Picone

etal.,

1989

(22)

RCT

n=43,Italy

Healthy,term

infants

n=10;experimentalformula

with

a50:50blendof

bovine

wheyandcasein

atdiffe

rent

concentra

-

tions

(1.12g/Lor

13.3g/L)for12

wk

n=10;formulawith

a

proteinconcentra

tion

of14.8g/L


head

circum

ferenceat

birth

and2,4,8,and12

wkof

age

NS

Studysupportedinpartby

Ross

Laboratories,

UnitedStates

Breastfedgroup,n=10

Age:fro

mbirth

Lonnerdaland

Chen,I

1990

(25)

RCT

n=notreporte

d,

China

Healthy,term

infants

n=notreported;fedexclu-

sivelyformulawith

the

followingproteinand

whey:caseinratio:

21.3g/dL,55:45

21.4g/dL,55:45

21.4g/dL,60:40

21.4g/dL,20:80;

67kcal/dLfor12

wk

n=notreporte

d;fed

exclusivelyformula

with

proteinandwhey:

caseinratio

1.5g/dL,

55:45

Meanweight,length,

weight:height,and

head

circum

ference

andskinfoldsat2,4,8,

and12

wkof

age

Similarfora

llgroups**

Studysupportedby

Wei-

ChuanFoods;Clinical

NutritionResearch

Unit

Five

formulagroups

and

oneexclusively

breastfedgroup

Age:fro

mbirth

Weightgain

Similarfora

llgroups**

Length

gain

Lonnerdaland

Chen,II1990

(26)

RCT

n=notreporte

d,

China

Healthy,term

infants,

formula-fed

frombirth

Follow-upformulawith

the

followingproteinand

whey:caseinratio:

21.5g/dL,55:45

with

25gcereal/d

assupple-

mentalfoodor

without

cereal

22.2g/dL,55:45

(67

kcal/dL)with

25gcereal/d

assupplementalfood

Commerciallyavailable

follow-upformula

(Promil)with

aprotein

contentof2.9g/dL

and

whey:caseinratio

60:40(67kcal/dL),w

ith

25gcereal/d

assup-

plem

entalfood

Meanweight,length,

weight:height,and

head

circum

ference

andskinfoldsat

5,6,

and7moof

age

Similarfora

llgroups**

Wei-Chuan

Foods;Clinical

NutritionResearch

Unit

(grant

no.P30-

AM235747)

Thesamepopulation

studiedinprevious

LonnerdalI

study

(0–4

mo)

Age:4mo

Weightgain

Similarfora

llgroups**

Authors�conclusions:

higher

proteinintake

appearsto

beexces-

sive;low

erproteinin-

take

isadequate

Length

gain

(Continued)



TABLE

2Continued

Reference

Stud

ytype

Samplesize

andsetting

Population

Lower-protein

group

Higher-protein

group

Outcomes

relevant

forthisreview

Effect

size

Funding

Comments

Szajew

skaet

al.,

1993

(18)

RCT

n=60,Poland

Healthy,term,

exclusively

formula-fed

infants

n=14;experimentalformula

with

1.5g/100mLpro-

tein;w

hey:caseinratio

60:40for12

wk

n=14;formulawith

a

proteincontentof

2.2

g/100mL;whey:casein

ratio

60:40;

n=15;formulawith

a

proteincontentof

2.2

g/100mL;whey:casein

ratio

18:82;

infant

formulacontain-

ing18

gprotein/L(n=

23at

3mo)


head

circum

ferenceat

day10;alsoat

weeks

4,8,and12

NS

Studysupportedinpartby

Milupa

Breastfedgroup,n=15

Age:fro

mbirth

Akeson

etal.,

1998

(20)

RCT

n=80,S

weden;

numberof

in-

fantsinthe

different

groups

changedover

time(see

Commentsrow)

Healthy,term

infants,

previously

exclusivelyBF

Infantformulacontaining

13

gprotein/L(n=25

at3

mo)or15

gprotein/L(n=

26at

3mo)gradually

intro

ducedto

infants

alongwith

breastfeeding.

Attheageof

6mo,

exclusivelyformula-fed

groups,a

mixed

feeding

group,andabreastfed

groupwereformed

Gainsinweightand

lengthat4–8and8–12

moof

age

NS

Studyfinanciallysup-

porte

dby

theSw

edish

NutritionFoundation

Infantsthat

withdrew

fromthestudywere

replaced

bynexteligi-

blesubject;at

6mo,

infantsfedformula

,125mL/dformed

the

breastfedgroup,and

thosefed.125mL/d

formed

themixed-

feedinggroup;then

got

backtoprevious

groups

whenbreastfeeding

was

reducedor

stop-

ped;Theseinfants

werereplaced

bythe

subjectsexclusively

formula-fed

Age:3mo

zscores

oftheabsolute

weightand

lengthat3,

6,and12

moof

age

NS

Crow

n–heellength,low

er

leglength,and

head

andarmcircum

fer-

ences

NS

Karlsland

Akeson

etal.,2000

(21)

RCT

n=80,S

weden;

n=59,Italy

Healthy,term

infants,

breastfed

for$2–3mo

Gradually

intro

duced(when

breastfeedingtermi-

nated)infant

formula

with

diffe

rent

protein

concentra

tions:

213

g/L;n=10

(Swedish

infants)andn=12

(Italian

infants)

215

g/L;n=9(Swedish

infants)andn=9(Italian

infants)exclusivelyfed

for$4mo(6–10moof

age;seeCommentscol-

umn);

inaddition,Sw

edishor

Mediterra

nean

weaning

foodsgiven

Infant

formulawith

pro-

teinconcentra

tions

of

18g/L(Swedishin-

fants;n=8)or

20g/L

(Italianinfants;n=9)

Weightand

lengthat3,4,

6,and12

moof

age

Normalgrow

th;nodiffe

r-

encesbetweencorre

-

sponding

Swedishand

Italiangroups

during

thestudyperiod

Studysupportedby

Ross

Laboratories,United

States;N

estle

Sweden

AB;the

AlbertPahlsson

Foundation,Sw

eden;

theF€ orenade

Liv

MutualG

roup

Life

InsuranceCompany,

Sweden;and

the

SwedishNutrition

Foundation

Anexclusioncriterionwas

breastfeedingat

$6

moof

age(47Sw

edish

and18

Italianinfants);

between10

and12

mo

ofage,formulaex-

changedforcow-milk

productsinmostItalian

infants;at

12moof

age,allS

wedishand

21%

(n=6of

28)of

Italianinfantsstillre-

ceived

formula

Age:3mo

Gainsinweightand

length

inSw

edishand

Italianinfantsbetween

3and6moand6and

12moof

age

Data

notcom

paredwithin

onegroup(low-vs.

high-proteinformulas)

**

(Continued)



TABLE

2Continued

Reference

Stud

ytype

Samplesize

andsetting

Population

Lower-protein

group

Higher-protein

group

Outcomes

relevant

forthisreview

Effect

size

Funding

Comments

Turcket

al.,2006

(27)

RCT

n=162,France

Healthyterm

infants

n=51;fed

exclusivelyiso-

caloric

whey-predom

i-

nant

(70:30

whey:casein

ratio)study

formula(pro-

tein:energyratio:1.8g/

100kcal;1.2g/100mL)

for120d

n=50;fed

exclusivelya

conventionalcasein-

predom

inant(70:30

casein:wheyratio)for-

mula(protein:energy

ratio:2.6g/100kcal;

1.7g/100mL)

Primaryoutcom

e:daily

weightgainbetween

days

0and120(non-

inferioritycriterion:dif-

ferenceindaily

weight

gain#4g);

Meandaily

weightgainin

theformula-fedgroups

diffe

redby

0.38

g/d

(95%

CI:2

2.59,1.83),

signifyingthenoninfer-

iorityof

thestudyfor-

mula**

Theisocaloricformulas

weresuppliedby

Nestle

France

Breastfed,referencegroup

(n=55)

Age:,7d

Secondaryoutcom

es:

daily

gaininweight,

length,headcircum

fer-

ence,and

BMIat

monthlyintervals

NS

CHOP

study(19,

30,31,32,*)

RCT

n=1138,m

ulti-

center

European

study

(Poland,Italy,

Germany,

Spain,

andBelgium)

Healthy,term

infants

n=540;1)infant

formula

with

1.25

gprotein/100

mL(1.77g/100kcal)and

energy

69.9g/100mL;

2)follow-onformulawith

aproteincontentof1.6g/

100mL(2.2g/100kcal)

andenergy

content72.7g/

100mL;afterintroduction

ofcomplem

entaryfeeding

(not

before

starto

fmonth

5);upto12

moof

age

n=550;1)infant

formula

with

aproteincontent

of2.05

g/100mL(2.9g/

100kcal);energy

69.8

g/100mL;

2)follow-onformula

with

aproteincontent

of3.2g/100mL(4.4g/

100kcal)and

energy

content72.7g/100mL;

afterintro

ductionof

complem

entaryfeeding

(not

before

startof

month

5);upto

12mo

ofage

Meanweightand

zscores

at3,6,12,24,and72

moof

age

Significantlyhigher

in

higher

proteinformula

grouponlyat

6and12

moof

age

Supportedby

EUgrants

(5th

and6thFP);the

InternationalD

anone

Institute;G

ermany:the

Child

HealthFoundation;

LMUinnovativeresearch

priorityproject;and

FederalM

inistryof

EducationandResearch

Breastfedgroupcreated;

additionalanalysesin

subsam

pleof

infants

done

(weightgain,

body

composition,

echocardiography,and

kidney

volume)

Age:fro

mbirth

to8wk

Meanlength

andzscores

at3,6,12,24,and72

moof

age

NS

Allformulas,w

hey:cas-

seinratio

1:4

Meanhead

circum

ference

andzscores

at6,12,

and24

moof

age

Not

reporte

d

MeanBM

Iand

BMIz

scores

at3,6,12,24,

and72

moof

age

Significantlyhigher

in

higher-proteinformula

groupfrom12

moofage

Prevalence

ofobesity

at6

yof

age

Significantlygreaterin

higher-proteinformula

group

Insulin

andIGF-Iaxis;

otherbiochemical

markersat

6moofage

Higher

proteinintake

stimulates

theIGF-I

axisandinsulin

release

1ForNS,differencebetw

eenthegroups,P.

0.05.*Unpublisheddata;**Resultsreportedbytheauthors

withoutprovidingnumericaldata

andstatisticalsignificance.BF,breastfed;CHOP,EUChildhoodObesityProgramme;ESPGHAN,European

Society

forPediatric

GastroenterologyHepatologyandNutrition;EU,EuropeanUnion;FP,framework

programme;IGF-I,insulin-likegrowth

factorI;LMU,Ludwig

Maxim

iliansUniversity;RCT,randomizedcontrolledtrial.



TABLE 3 Results of individual studies (not included in a meta-analysis) evaluating the effects of varyingprotein concentrations in infant formulas on growth markers and BMI, assessed at different ages ofchildren1

Outcome, reference, and age of childrenLower

protein, nHigher

protein, nStatisticalmethod

Effect size(95% CI)

Obesity

Weber et al., 2014 (30), 6 y 227 221 RD 20.06 (20.10, 20.01)

Mean BMI, kg/m2

CHOP study (*), 3 mo 454 451 MD 20.07 (20.25, 0.10)

Turck et al., 2006 (27), 4 mo 51 50 MD 0.24 (20.24, 0.72)

CHOP study (*), 6 mo 419 417 MD 20.29 (20.49, 20.09)

CHOP study (*), 12 mo 374 374 MD 20.33 (20.55, 20.11)

Koletzko et al., 2009 (19), 24 mo 313 322 MD 20.30 (20.49, 20.11)

Weber et al., 2014 (30), 6 y 227 221 MD 20.50 (20.89, 20.11)

BMI z score

CHOP study (*), 3 mo 454 451 MD 20.04 (20.16, 0.08)

CHOP study (*), 6 mo 419 417 MD 0.11 (20.02, 0.24)

CHOP study (*), 12 mo 374 374 MD 20.23 (20.38, 20.08)


Weber et al., 2014 (30), 6 y 227 221 MD 20.30 (20.52, 20.08)

Mean weight, kg

Zoppi et al., 1978 (29), 5.5 mo 6 7 MD 20.96 (21.71, 20.22)

CHOP study (*), 6 mo 419 417 MD 20.16 (20.27, 20.04)

Zoppi et al., 1978 (29), 10 mo 9 7 MD 20.28 (21.59, 1.03)

CHOP study (*), 12 mo 374 374 MD 20.24 (20.40, 20.09)


Weber et al., 2014 (30), 6 y 227 221 MD 20.59 (21.31, 0.13)

Mean weight z score

CHOP study (*), 3 mo 454 451 MD 20.09 (20.19, 0.02)

CHOP study (*), 6 mo 419 417 MD 20.16 (20.28, 20.04)

CHOP study (*), 12 mo 374 374 MD 20.20 (20.33, 20.07)


Weber et al., 2014 (30), 6 y 227 221 MD 20.15 (20.36, 0.06)

Mean weight gain, g/d

R€aih€a et al., 1986 (24), 2 wk to 3 mo 10 10 MD 21.57 (27.98, 4.84)

Turck et al., 2006 (27), 0–4 mo 36 38 MD 0.40 (21.72, 2.52)

Zoppi et al., 1978 (29), 0–5.5 mo 6 7 MD 22.40 (26.07, 1.27)

Zoppi et al., 1978 (29), 0–10 mo 9 7 MD 21.90 (26.41, 2.61)

Akeson et al., 1998 (20), 4–8 mo 51 23 MD 0.92 (20.99, 2.83)

Akeson et al., 1998 (20), 8–12 mo 51 23 MD 20.79 (22.18, 0.60)

Mean weight gain, g/(kg � d)Akeson et al., 1998 (20), 4–8 mo 51 23 MD 0.17 (20.18, 0.52)

Akeson et al., 1998 (20), 8–12 mo 51 23 MD 20.09 (20.25, 0.07)

Mean length:height, cm

CHOP study (*), 6 mo 419 417 MD 20.10 (20.40, 0.20)

CHOP study (*), 12 mo 374 374 MD 20.20 (20.58, 0.18)


Weber et al., 2014 (30), 6 y 227 221 MD 0.39 (20.47, 1.25)

z score for absolute length:height

CHOP study (*), 3 mo 454 451 MD 20.09 (20.22, 0.04)

CHOP study (*), 6 mo 419 417 MD 20.03 (20.16, 0.10)

CHOP study (*), 12 mo 374 374 MD 20.08 (20.23, 0.07)


Weber et al., 2014 (30), 6 y 227 221 MD 0.07 (20.10, 0.24)

Mean length gain, cm/mo

Zoppi et al., 1978 (29), 0–5.5 mo 6 7 MD 20.10 (20.48, 0.28)

Zoppi et al., 1978 (29), 0–10 mo 9 7 MD 0.10 (20.34, 0.54)

Mean length gain, mm/(m � d)Akeson et al., 1998 (20), 4–8 mo 51 23 MD 0.00 (20.07, 0.07)

Akeson et al., 1998 (20), 8–12 mo 51 23 MD 20.02 (20.08, 0.04)

(Continued)



The risk of bias in included studies

The included studies are described with respect to their risk ofbias in Supplemental Table 3. Methodologic limitationsincluded a very small sample size, participants� replacement,pseudo-random allocation, a high loss to follow-up, per-protocol analysis, and selective outcomes reporting. In manystudies, because of poor reporting, unambiguous evaluation ofthe risk of bias was not possible, which often resulted in thedescription of some elements as unclear.

Effects of interventions—primary outcomes

Overweight and obesity. Only one RCT (30) (the CHOPstudy; n = 448) reported this outcome (Table 3). The risk ofbecoming obese at the age of 6 y was significantly lower in thelower-protein formula group than in the higher-protein formulagroup (RR 0.44; 95% CI: 0.21, 0.91).

Fat mass and fat-free mass. The analysis of body compositionby isotope dilution was performed in a subsample (n = 52) of 6-mo-old infants who were enrolled in the CHOP study (31).Although correlations between the fat-mass z score (but not thefat-free mass z score) and both growth velocity and BMI wereobserved, fat-free mass and fat mass z scores did not differbetween the higher- and lower-protein formula groups. In thestudy by Lonnerdal and Chen (26), skin fold measurements wereobtained at 5, 6, and 7 mo of age. The authors reported similar

results for all study groups. However, no numeral data wereavailable.

Anthropometric growth markers.BMI. This outcome was reported at different time points in 4

publications (19, 27, 30, 32) (Table 3). At 3 or 4 mo of age, therewere no significant differences between the study groups.Starting at 6 mo of age, only one RCT (CHOP Study) reporteddata on BMI. BMI was significantly lower in the lower-proteinformula group than in the higher-protein formula group at alltime points (i.e., at 6, 12, and 24 mo and at 6 y).

BMI z scores. Only the CHOP study (unpublished data)reported BMI z scores (Table 3). No significant differences werefound at 3 and 6 mo of age between the lower-protein formulagroup and the higher-protein formula group. Starting at 12 moof age, BMI z scores were significantly lower in the lower-proteinformula group than in the higher-protein formula group at alltime points (i.e., at 12 and 24 mo and at 6 y).

Mean weight. Twelve publications (18–20, 22–26, 29–32)reported mean weight at different time points (Figure 1 andTable 3). The pooled results of 4 RCTs (18, 19, 22, 24) showedno significant difference between the groups consuming lower-protein and higher-protein formula at 3 mo of age (n = 998;MD,–0.07 kg; 95% CI: 20.15, 0.01). In one small (n = 13) trial byZoppi et al. (29), mean weight was significantly lower in infantsfed formula with a lower-protein concentration than in those fed

TABLE 3 Continued

Outcome, reference, and age of childrenLower

protein, nHigher

protein, nStatisticalmethod

Effect size(95% CI)

Mean length gain, mm/d

Akeson et al., 1998 (20), 4–8 mo 51 23 MD 20.01 (-0.05, 0.03)

Akeson et al., 1998 (20), 8–12 mo 51 23 MD 20.03 (20.07, 0.01)

Mean crown–rump length gain, cm/wk


Mean crown–heel length gain, cm/wk


Weight-for-length z score

CHOP study (*), 3 mo 454 451 MD 20.02 (20.15, 0.11)

CHOP study (*), 6 mo 419 417 MD 20.19 (20.32, 20.06)

CHOP study (*), 12 mo 374 374 MD 20.23 (20.38, 20.08)


1 *Unpublished data. CHOP, EU Childhood Obesity Programme; MD, mean difference; RD, risk difference.

FIGURE 1 Forest plot of the effects of varying protein concentrations in infant formulas on the mean weight of infants at 3 mo of age.

*Unpublished data; ^combined results from 2 study arms. CHOP, EU Childhood Obesity Programme; F2.2, study group with formula

protein concentration of 2.2 g/100 mL; F11, study group with formula protein concentration of 11.2 g/L; F13, study group with formula protein

concentration of 13.3 g/L; IV, inverse variance.



formula with a higher-protein concentration at 5.5 mo of age;however, the difference between the groups was no longersignificant when these infants were 10 mo old. The CHOP studyreported a significantly lower mean weight in the group consum-ing lower-protein formula than in the group consuming higher-protein formula at 6 and 12mo of age, but there was no significantdifference between the groups at 24 mo and at 6 y.

In one study (23), this outcome was assessed at different childages in the groups consuming lower-protein and higher-proteinformula. Therefore, we found it improper to compare the results.

Mean weight z scores. Only the CHOP study (unpub-lished data) reported mean weight z scores (Table 3). Meanweight z scores were significantly lower at 6 and 12 mo of agein the group consuming the lower-protein formula than in thegroup consuming the higher-protein formula; however, therewere no significant differences between the groups at 3 mo,24 mo, and 6 y of age.

Mean weight gain. The authors of almost all of the studies[except for the studies by Szajewska et al. (18), Picone et al. (22),and the CHOP study, unpublished data] reported gains in weightin a variety of ways (e.g., different units or different timeintervals) (Figure 2 and Table 3). However, in almost all of thecomparisons (regardless of the time intervals assessed), as well aswhen the results of some studies were pooled together into a

meta-analysis, no statistically significant difference between thegroups consuming lower-protein and higher-protein formulaswas found in infants up to 12 mo of age.

Mean length/height. Mean length/height (absolute meanlength/height or the z score for absolute length/height) wasreported in 7 RCTs (18–20, 22, 24–26) (Figure 3 and Table 3).The pooled results of 4 RCTs (18, 19, 22, 24) revealed asignificantly shorter length at 3 mo of age in the group consuminga lower-protein formula than in the group consuming the higher-protein formula (n = 998; MD, –0.27 cm; 95% CI: 20.52,20.02). No significant differences between groups were found atany other time points (i.e., at 6 mo, 12 mo, and 24 mo, and 6 yof age). Three RCTs (20, 25, 26) provided only descriptiveresults (concluding that length was similar in all groups).

Length gain. Length gain up to 12 mo of age was assessedin 10 RCTs (20–29) (Figure 4 and Table 3). Different units anddifferent approaches to length measurements (i.e., length,crown–heel length, and crown–rump length) were used toexpress this outcome. The majority of available results showed asimilar length gain in infants regardless of the protein concen-tration in the consumed formula.

In 2 studies performed by Lonnerdal and Chen (25, 26), itwas not clearly stated whether the authors described a similarmean weight and length or weight and length gain.

FIGURE 2 Forest plot of the effects of varying protein concentrations in infant formulas on the mean weight of infants from 3 to 6 mo and from

6 to 12 mo of age. ^Combined results from 2 study arms. F13, study group with formula protein concentration of 13 g/L; F15, study group with

formula protein concentration of 15 g/L; IV, inverse variance.

FIGURE 3 Forest plot of the effects of varying protein concentrations in infant formulas on the mean length/height of infants at 3 mo of age.

*Unpublished data; ^combined results from 2 study arms. CHOP, EU Childhood Obesity Programme; F2.2, study group with formula

protein concentration of 2.2 g/100 mL; F11, study group with formula protein concentration of 11.2 g/L; F13, study group with formula protein

concentration of 13.3 g/L; IV, inverse variance.



Head circumference. In 5 RCTs (18, 22, 24–26), dataregarding head circumference were available (Figure 5). Thepooled results of 3 RCTs (18, 22, 24) showed no significantdifference in head circumference between the groups consuminglower-protein and higher-protein formula at 12 wk of age (MD,20.31 cm; 95% CI: 20.63, 0.02). However, significant heteroge-neity was found (I2 = 81%). The observed heterogeneity was notexplained by the subgroup analysis based on the amount of proteinin milk formulas, which was performed as originally planned inthe protocol. We also considered different measurement techniquesas a potential source of the identified heterogeneity. However, theauthors did not report how the measurements were obtained. In2 studies by Lonnerdal and Chen (25, 26), the authors reported asimilar mean head circumference in all study groups (numericaldata were not shown).

Head circumference gain. In the study by Turck et al. (27),the daily gain in head circumference (millimeters per day) atmonthly intervals was described, showing a similar growth rate inboth study groups. In the study by Zoppi et al. (23), the meanincrease in head circumference was evaluated in the group consum-ing higher-protein formula only. In another report (28), the meandaily gain in head circumference was described as normal.

Weight-for-length z scores. Weight-for-length z scoreswere analyzed in the CHOP study [(unpublished data) Table 3].The weight-for-length z scores were significantly higher at 6, 12,and 24mo of age in the group consuming higher-protein formulathan in the group consuming lower-protein formula.

Other. Other anthropometric assessments were reported insome of the included trials (particularly an early assessment ofgrowth at 2, 4, and 8 wk of age; evaluation of weight and length

gain at monthly intervals; etc.). However, we found the results ofthese trials to be irrelevant to our review.

Secondary outcomes

Our secondary outcomes were only partially addressed in 2studies. A summary of these findings is presented in Table 4.

Discussion

Summary of the evidence. This review revealed that the vastmajority of identified studies evaluated only short-term out-comes related to infant growth. The gathered evidence suggeststhat during the first year of life, different protein concentrationsin infant formulas do not seem to affect infants� linear growthsignificantly apart from a transient effect on mean length duringthe first months of life. Data with respect to weight and weightgain suggest similar growth in the first months of life, with alower mean weight (and weight z score) obtained in the infantsfed lower-protein formulas from;6 mo until 12 mo of age only.A significantly lower BMI associated with the intake of lower-protein formula was firmly observed in 12-mo-old infants and inchildren up until early school age. No conclusions with regard tothe effects of formula protein concentration on body composi-tion can be formulated. We have identified only one RCT (theCHOP study) that assessed the outcomes of interest afterinfancy. The intake of lower-protein formula during infancy wasobserved to reduce the risk of obesity at the age of 6 y in thisstudy. However, these findings need to be interpreted withcaution, because this is a single report, and some methodologiclimitations of the study were identified.

FIGURE 4 Forest plot of the effects of varying protein concentrations in infant formulas on the mean length gain of infants from 3 to 6 mo and

from 6 to 12 mo of age. ^Combined results from 2 study arms. F13, study group with formula protein concentration of 13 g/L; F15, study group

with formula protein concentration of 15 g/L; IV, inverse variance.

FIGURE 5 Forest plot of the effects of varying protein concentrations in infant formulas on the mean head circumference of infants at 3 mo of

age. ^Combined results from 2 study arms. F2.2, study group with formula protein concentration of 2.2 g/100 mL; F11, study group with formula

protein concentration of 11.2 g/L; F13, study group with formula protein concentration of 13.3 g/L; IV, inverse variance.



Strengths and limitations. The major strength of our review isthat it collates the largest number of studies, to our knowledge,available on the effects of varying protein concentrations ininfant formulas on growth, body composition, and the later riskof obesity. We used a rigorous systematic review methodologyproposed by the Cochrane Collaboration. We used severalmethods to reduce bias from the initial preparatory stageuntil the final report (i.e., protocol registration, comprehen-sive literature search, duplicate data abstraction, and prespecifiedcriteria for methodologic assessment and analysis). We did notimpose restrictions by language or year of publication. Attemptswere made to identify unpublished trials. However, this review hassome limitations. The methodologic quality of the included trialsvaried. Potential limitations include unclear sequence generation,unclear allocation concealment and blinding, small samplesizes, lack of sample size calculations, and high dropout rates insome trials, including the CHOP trial. Despite the fact that anumber of trials met the inclusion criteria, our conclusions afterthe infancy period are based solely on the results of one RCT (theonly experimental study with a large sample size and longerfollow-up). It is worth mentioning that the sample sizes of theincluded studies varied, and that sample size in the CHOP studywas greater than the total number of participants of all otherincluded studies. We observed a great deal of heterogeneity inthe protein concentrations in the formulas used in differentstudies, with the protein amount being the same in groupsconsuming lower- and higher-protein formula, depending on thestudy in some cases. Therefore, our review cannot provide ananswer to the question of what the particular and optimalamount of protein to be included in infant formulas should be.

Another important aspect raised by our review is that notonly protein quantity, but also protein quality, must be takeninto account to secure a sufficient amount of essential aminoacids to support adequate growth and safety in infants. Stud-ies that use the Indicator Amino Acid Oxidation method todetermine the actual requirement of each essential amino acid inneonates have shown that infant formulas that are currentlyused may not contain the optimal amino acid composition andmay provide an amount of essential amino acids that is too high(35–37). However, a detailed analysis of the amino acid contentof formula was outside the scope of our review. Also, we haveexcluded from our analysis a number of studies in whichadditional differences between the groups were present, such asenriched milk formula (with a-lactalbumin, for example) orformula with extra components added (see exclusion criteria),apart from quantitative protein modifications, differences inthe whey-to-casein ratio, and other differences in composition

secondary to the protein amount changes (in order to achieve anisocaloric formula). These restrictive criteria were developed toreduce the great heterogeneity between the included studies.However, we realize that these data may be a substantialaddition to this complex issue. Finally, we analyzed only oneaspect of protein intake provided by milk formula consumption,whereas other sources of protein in the diet also play animportant role after the introduction of complementary feeding.

Agreement and disagreement with other studies orreviews. The findings of this review are in accordance with 2recently published reviews. The first review by Hornell et al.(38), published in 2013, focused on protein intake in the diet(not only in milk formula) of children of different ages (0–18 y)and its relation to health based on both experimental andobservational studies. The authors concluded that higherprotein intake during infancy is associated with increasedgrowth, a higher BMI during childhood, and an increased riskof being overweight later in life. The second review by Abramset al. (39), published when this manuscript was in the final stagesof preparation, addressed the effect of the intake of low-proteinand low-energy formula by full-term infants on growth. Sixpublications were included in that review. Based on the findingsfrom 4 RCTs, the authors concluded that the use of infantformulas with concentrations of protein and energy slightlybelow historical standards in the United States results inadequate growth during infancy and early childhood. Theauthors also stated that ‘‘Further long-term research is needed toassess the impact of the use of lower-protein and/or lower-energyproducts, especially for nutritionally at-risk populations such aspreterm infants and infants who are born small forgestational age’’ (39). Although the authors asked a similarquestion to the one that is the subject of our review, theirapproach differed. The search was limited to 3 electronicdatabases only (i.e., Medline, Cochrane, and the CumulativeIndex to Nursing and Allied Health Literature). Furthermore,the set of key words used for searching differed, as well as theinclusion and exclusion criteria, resulting in the differencesobserved in the number of identified trials (4 studies in thereview by Abrams et al. (39) compared with 12 trials identifiedin this review). This was not unexpected, because it is wellknown that small differences in search strategies may result insubstantial differences in the set of trials identified. Nevertheless,taken together, all 3 reviews clearly prove how complex the issueof protein intake can be and leave room for more evidence andother systematic analyses to answer the questions that remainunanswered.

TABLE 4 Effects of different protein concentrations in infant formulas on lipid profile, IGF-I axis, and kidney size assessed duringinfancy1

Reference Outcome Results

Zoppi et al. (23) Serum cholesterol and TG concentrations at 5 mo of age Observed values largely within normal range

Concentrations reflected amount of protein consumed

No statistically significant difference between the study

groups observed

CHOP (32) IGF-I axis Higher protein intake stimulates IGF-I axis and insulin release

CHOP (33, 34) Kidney size (at 6 mo of age) Significantly increased kidney size in infants who consumed

a higher-protein formula

IGF-I in part mediates protein-induced kidney growth

Effect of higher protein intake during early infancy on long-term

kidney function requires further evaluation

1 CHOP, EU Childhood Obesity Programme; IGF-I insulin-like growth factor I.



The evidence is insufficient for assessing the effects ofreducing protein concentration in infant formulas on long-term outcomes, but, if confirmed, this could represent apromising intervention for reducing the risk of overweight andobesity in children. In view of the limited available evidence,more studies replicating the effects on long-term health out-comes are needed. Careful analysis of observational studies thathave assessed long-term outcomes of interest may provide avaluable addition to the limited experimental evidence.

Acknowledgments

BK, JBvG, and HS initially conceptualized this study; BP-G,BMZ, SMPK, and JK were responsible for data collection, dataanalysis, data interpretation, and preparation of the report; BP-Gassumed primary responsibility for the writing of the first draft ofthe manuscript; and all authors contributed to the initial protocolof the study. All authors read and approved the final manuscript.

References

1. World Health Organization. Protein and amino acid requirements inhuman nutrition. [Technical Report Series, 2002 935:265.]

2. Inostroza J, Haschke F, Steenhout P, Grathwohl D, Nelson S, Ziegler E.Low-protein formula slows weight gain in infants of overweightmothers. J Pediatr Gastroenterol Nutr 2014;59:70–7.

3. Toschke AM, Grote V, Koletzko B, von Kries R. Identifying children athigh risk for overweight at school entry by weight gain during the first 2years. Arch Pediatr Adolesc Med 2004;158:449–52.

4. Brands B, Demmelmair H, Koletzko B. EarlyNutrition Project. Howgrowth due to infant nutrition influences obesity and later disease risk.Acta Paediatr 2014;103:578–85.

5. Koletzko B. Early nutrition and its later consequences: new opportu-nities. In: Koletzko B, Akerblom H, Dodds PF, Ashwell M (eds). Earlynutrition and its later consequences: new opportunities. Adv Exp MedBiol 2005;569:1–12.

6. Rolland-Cachera MF, Deheeger M, Akrout M, Bellisle F. Influence ofmacronutrients on adiposity development: a follow up study ofnutrition and growth from 10 months to 8 years of age. Int J ObesRelat Metab Disord 1995;19:573–8.

7. Kirchberg FF, Harder U, Weber M, Grote V, Demmelmair H, PeissnerW, Rzehak P, Xhonneux A, Carlier C, Ferre N, et al.; EuropeanChildhood Obesity Trial Study Group. Dietary protein intake affectsamino acid and acylcarnitine metabolism in infants aged 6 months. JClin Endocrinol Metab 2015;100:149–58.

8. Tijhuis MJ, Doets EL, Noordegraaf-Schouten V. Extensive literaturesearch and review as preparatory work for the evaluation of theessential composition of infant and follow-on formulae and growing-upmilk [Internet]. [cited 2015 Apr 20.] Available from: www.efsa.europa.eu/en/supporting/doc/551e.pdf.

9. Koletzko B. Standards for infant formula milk. Commercial inter-ests may be the strongest driver of what goes into formula milk. BMJ2006;332:621–2.

10. Koletzko B, Baker S, Cleghorn G, Neto UF, Gopalan S, Hernell O, HockQS, Jirapinyo P, Lonnerdal B, Pencharz P, et al. Global standard for thecomposition of infant formula: recommendations of an ESPGHANcoordinated international expert group. J Pediatr Gastroenterol Nutr2005;41:584–99.

11. European Union Commission Directive 2006/141/EC. Foods for infants& young children - Infant formulae and follow-on formulae. [Internet].[cited 2015 Mar 20.] Available from: ec.europa.eu/food/food/label-lingnutrition/children/formulae_en.htm.

12. US Food and Drug Administration. Current manufacturing regulationsand guidelines. 21 CFR \107.100 Nutrient requirements for infantformulas [Internet.] [cited 2015 Mar 20.] Available from: www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInformation/InfantFormula/default.htm.

13. European Union Commission Directive 2006/141/EC. Infant formulaeand follow-on formulae. Official Journal of the European Communi-ties [Internet]. [cited 2015 Mar 20.] Available from: eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ%3AL%3A2006%3A401%3A0001%3A0033%3AEN%3APDF.

14. US Food and Drug Administration. Food and Drug Administration rulesand regulations. Nutrient requirements for infant formulas (21 CFR,Part 107). Code of Federal Regulations. Title 21, Volume 2. Revised asof April 1, 2014. 21CFR107.100.

15. Codex Alimentarius Commission. Standard for infant formulaand formulas for special medical purposes intended for infants(Codex Stan 72–1981) [Internet]. [cited 2015 Mar 20.] Availablefrom: www.codexalimentarius.org/input/download/standards/288/CXS_072e.pdf.

16. European Food Safety Agency, Panel on Dietetic Products, Nutritionand Allergies. Scientific opinion on the essential composition of infantand follow-on formulae. EFSA Journal. 2014;12:3760.

17. Higgins JPT, Green S. (eds)Cochrane Handbook for systematic reviews ofinterventions version 5.1.0 [Internet]. The Cochrane Collaboration,2011. [cited 2014 Jul 21.] Available from: www.cochrane-handbook.org.

18. Szajewska H, Albrecht P, Krawczyk A. The effect of various intakes ongrowth and protein metabolizm of term infants during the first threemonths of life (part I). Pediatr Pol 1993;LXVIII:11.

19. Koletzko B, von Kries R, Closa R, Escribano J, Scaglioni S, GiovanniniM, Beyer J, Demmelmair H, Gruszfeld D, Dobrzanska A, et al.;European Childhood Obesity Trial Study Group. Lower protein ininfant formula is associated with lower weight up to age 2 y: arandomized clinical trial. Am J Clin Nutr 2009;89:1836–45.

20. Akeson PM, Axelsson IE, Raiha NC. Growth and nutrient intake inthree- to twelve-month-old infants fed human milk or formulas withvarying protein concentrations. J Pediatr Gastroenterol Nutr1998;26:1–8.

21. Karlsland Akeson PK, Axelsson IE, Raiha NC, Warm A, Minoli I, MoroG. Protein intake and metabolism in formula-fed infants given Swedishor Italian weaning foods. Acta Paediatr 2000;89:158–64.

22. Picone TA, Benson JD, Moro G, Minoli I, Fulconis F, Rassin DK, RaihaNC. Growth, serum biochemistries, and amino acids of term infants fedformulas with amino acid and protein concentrations similar to humanmilk. J Pediatr Gastroenterol Nutr 1989;9:351–60.

23. Zoppi G, Gerosa F, Pezzini A, Bassani N, Rizzotti P, Bellini P, TodeschiniG, Zamboni G, Vazzoler G, Tridente G. Immunocompetence anddietary protein intake in early infancy. J Pediatr Gastroenterol Nutr1982;1:175–82.

24. Raiha N, Minoli I, Moro G. Milk protein intake in the term infant. I.Metabolic responses and effects on growth. Acta Paediatr Scand1986;75:881–6.

25. Lonnerdal B, Chen CL. Effects of formula protein level and ratio oninfant growth, plasma amino acids and serum trace elements. I. Cow�smilk formula. Acta Paediatr Scand 1990;79:257–65.

26. Lonnerdal B, Chen CL. Effects of formula protein level and ratio oninfant growth, plasma amino acids and serum trace elements. II. Follow-up formula. Acta Paediatr Scand 1990;79:266–73.

27. Turck D, Grillon C, Lachambre E, Robiliard P, Beck L, Maurin JL,Kempf C, Bernet JP, Marx J, Lebrun F, et al. Adequacy and safety of aninfant formula with a protein/energy ratio of 1.8 g/100 kcal andenhanced protein efficiency for term infants during the first 4 months oflife. J Pediatr Gastroenterol Nutr 2006;43:364–71.

28. Zoppi G, Gasparini R, Mantovanelli F, Gobio-Casali L, Astolfi R,Crovari P. Diet and antibody response to vaccinations in healthy infants.Lancet 1983;2:11–4.

29. Zoppi G, Zamboni G, Siviero M, Bellini P, Cancellieri ML. Gamma-Globulin level and dietary protein intake during the first year of life.Pediatrics 1978;62:1010–8.

30. Weber M, Grote V, Closa-Monasterolo R, Escribano J, Langhendries JP,Dain E, Giovannini M, Verduci E, Gruszfeld D, Socha P., Koletzko forThe European Childhood Obesity Trial Study Group. Lower protein ininfant formula reduces BMI and obesity risk at school age: follow-up ofa randomized trial. Am J Clin Nutr 2014;99:1041–51.

31. Escribano J, Luque V, Ferre N, Mendez-Riera G, Koletzko B, Grote V,Demmelmair H, Bluck L, Wright A, Closa-Monasterolo R.; EuropeanChildhood Obesity Trial Study Group. Effect of protein intake andweight gain velocity on body fat mass at 6 months of age: the EUChildhood Obesity Programme. Int J Obes (Lond) 2012;36:548–53.

32. Socha P, Grote V, Gruszfeld D, Janas R, Demmelmair H, Closa-Monasterolo R, Subıas JE, Scaglioni S, Verduci E, Dain E, et al.;European Childhood Obesity Trial Study Group. Milk protein intake, themetabolic-endocrine response, and growth in infancy: data from arandomized clinical trial. Am J Clin Nutr 2011; 94(6, Suppl)1776S–84S.



33. Luque V, Escribano J, Grote V, Ferre N, Koletzko B, Gruszfeld D,

Socha P, Langhendries JP, Goyens P, Closa-Monasterolo R.; Euro-

pean Childhood Obesity Project. Does insulin-like growth factor-

1 mediate protein-induced kidney growth in infants? A secondary

analysis from a randomized controlled trial. Pediatr Res 2013;74:

223–9.

34. Escribano J, Luque V, Ferre N, Zaragoza-Jordana M, Grote V, Koletzko

B, Gruszfeld D, Socha P, Dain E, Van Hees JN, et al. Increased protein

intake augments kidney volume and function in healthy infants. Kidney

Int 2011;79:783–90.

35. de Groof F, Huang L, van Vliet I, Voortman GJ, Schierbeek H,

Roksnoer LC, Vermes A, Chen C, Huang Y, van Goudoever JB.

Branched-chain amino acid requirements for enterally fed term neonates

in the first month of life. Am J Clin Nutr 2014;99:62–70.

36. Huang L, Hogewind-Schoonenboom JE, de Groof F, Twisk JW,

Voortman GJ, Dorst K, Schierbeek H, Boehm G, Huang Y, Chen C,

et al. Lysine requirement of the enterally fed term infant in the first

month of life. Am J Clin Nutr 2011;94:1496–503.

37. Huang L, Hogewind-Schoonenboom JE, van Dongen MJ, de Groof F,

Voortman GJ, Schierbeek H, Twisk JW, Vermes A, Chen C, Huang Y, et al.

Methionine requirement of the enterally fed term infant in the first month

of life in the presence of cysteine. Am J Clin Nutr 2012;95:1048–54.

38. Hornell A, Lagstrom H, Lande B, Thorsdottir I. Protein intake from 0 to

18 years of age and its relation to health: a systematic literature review for

the 5th Nordic Nutrition Recommendations. Food Nutr Res 2013;57:.

39. Abrams SA, Hawthorne KM, Pammi M. A systematic review of

controlled trials of lower-protein or energy-containing infant formulas

for use by healthy full-term infants. Adv Nutr 2015;6:178–88.

