Hydrolysed formula and risk of allergic or autoimmune disease: systematic review and meta-analysis
BMJ 2016; 352 doi: https://doi.org/10.1136/bmj.i974 (Published 08 March 2016) Cite this as: BMJ 2016;352:i974
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Dear editor,
We read the recent publication of the systematic review and meta-analysis on effects of hydrolysed infant formula on the risk of allergic or autoimmune disease with great interest [1].
The authors conclude that there is no evidence that the feeding of infant formula based on partially or extensively hydrolysed protein reduces the risk of allergy or type 1 diabetes in infants with a positive family history. The authors propose a high or unclear risk of bias in most studies, and some indication for publication bias for studies on eczema and wheeze.
The validity of the authors’ conclusion is questioned by several limitations, which we wish to point out here.
1. In their meta-analysis, the authors pool the results for very different protein hydrolysates based on an assumed similar degree of hydrolysis. This is inappropriate because different biological effects of various hydrolysates are not only based on molecular mass distribution, but also on different peptide characteristics and sequence profiles which are quite consistent among different batches of the same product [2]. For example, the GINI study demonstrated that neither the degree of hydrolysation, nor in vitro molecular mass distribution of remaining peptide size or the protein source (casein or whey) predicted clinical outcomes [3-7]. In analogy to intervention trials with so called “probiotic” bacteria, where it is inappropriate to pool studies using different bacterial strains in one meta-analysis, it is inappropriate to mix “apples and oranges” of very different hydrolysate formulae interventions in a pooled meta-analysis, which may mask detection of effective interventions. Therefore, the European Food Safety Authority (EFSA) never accepted claims for preventive efficacy for the whole group of hydrolysed formulae, but supported separate evaluation of each particular product.
2. The authors’ meta-analysis ignores considerable differences in trial quality across the included studies. For example, they classify the study by Lowe et al. (third largest weight in the meta-analysis) as low risk of assessment bias (supplement table C) although blinding of study personal was not guaranteed [8]. Pooling studies with variable high to poor quality standards carries a high risk of producing misleading results, because the overall effect estimates might be strongly influenced by the sample size of a poor quality study. Therefore, regulatory bodies such as the US FDA and the European EFSA would not deduct recommendation on a pooled analysis of studies of good and poor quality, but base their conclusions on the use of partially hydrolysed formula in infants at risk for allergy on the results of high quality studies.
3. This meta-analysis cited the results of the GINI study incorrectly. In the meta-analysis for prevention of eczema in children <4 years (table 4) the results of von Berg et al. 2007 [4] were included but not cited, while a previous reference is inappropriately cited [3]. Of importance, Boyle et al. include the results of more than 800 exclusively breast-fed infants who participated in the GINI study but never received any study formula. The data of these infants do not contribute to addressing the question whether any of the different hydrolysed formulae reduces the risk of allergy, as compared to a cows’ milk protein formula. In the GINI study, infants were randomized at birth. The study team strongly supported breastfeeding, and accordingly 39.5% of infants were exclusively breastfed for the full 4 months of the planned intervention period. However, exclusively breastfed infants differed significantly from formula fed infants with respect to factors related to allergy risk, such as maternal smoking habits, furry pet keeping at home, solid food introduction, parental educational level, and study region [9].
The recently published results of the PATCH trial [10] on a partially hydrolysed whey based formula combined with added prebiotics support our findings of an association between breastfeeding and allergy risk factors. These investigators enrolled eligible infants at high risk for allergy (n=1047) after birth and followed them all, but randomized them only when the parents decided to start formula feeding. Only 18 % (n=184) of the infants in the PATCH-Study were exclusively breastfed during the first 4 months of age, compared to 39.5% in GINI. Exclusively breastfed infants in the PATCH study also showed striking differences from infants exposed to randomized formulae, with a more frequent family history of allergy in both parents (30% versus 20%), and a higher cumulative incidence of eczema until 12 months of age (44.2% versus 30.8 in the hydrolysed formula group and 30.2% in the control group, respectively). It would be inappropriate to conclude that breastfeeding causes an increased risk of eczema based on this self-selected group carrying higher risk factors. Boyle et al included in their meta-analysis the GINI data of ITT with all exclusively breastfed infants, which is not informative regarding the question raised by the authors. In addition, Boyle used the ITT results at 3 years based on clinical examination. However, these ITT results included only part of exclusively breastfed infants of one study region, because for financial reasons we could not offer clinical examination to all breastfed children in the second and third year of life increasing further the risk of selection bias. However, results of parent reported outcome based on doctor diagnosed eczema were available in the total cohort. We consider ITT analyses important, but in the case of the GINI study the strong support of breastfeeding resulted in a very high proportion of exclusively breastfed infants in the ITT group that were never exposed to the intervention, therefore the ITT analysis excluding these breastfed infants and the PP analyses must not be ignored.
Table 1 summarizes the results of the ITT analysis, including exclusively breastfed infants, the ITT analysis with exclusion of exclusively breastfed infants, and the PP analysis, with data on eczema, the dominant atopic manifestation in the GINI study, for the first 3 years of follow up. Regarding the three analysis population the results clearly demonstrate a significant and clinically meaningful reduction of eczema risk until 3 years for 2 of the 3 study formulae, as compared to cow’s milk formula.
4. The authors inappropriately classify the GINI study as “unclear” with respect to a potential conflict of interest [1]. Previously it has been transparently reported that the GINI study was financed until the three year follow up by public funding from the Germany Federal Ministry for Education, Science, Research, and Technology (grant no. 01 EE 9401-4) [4]. Four different manufacturers donated their formulae free of charge under the public grant agreement, which was necessary to ensure packaging in identical cans for full blinding. None of the manufacturers of the study formulae had any influence on study design, the data analysis, and interpretation of the results.
We conclude that the meta-analysis published by Boyle et al has serious methodological limitations, and the authors’ conclusions are misleading. Their meta-analysis does not allow conclusions on allergy preventive effects of different hydrolysed formulae in infants at increased risk for allergies.
Literature
1 Boyle RJ, Ierodiakonou D, Khan T, Chivinge J, Robinson Z, Geoghegan N, Jarrold K, Afxentiou T, Reeves T, Cunha S, Trivella M, Garcia-Larsen V, Leonardi-Bee J: Hydrolysed formula and risk of allergic or autoimmune disease: systematic review and meta-analysis. BMJ 2016;352:i974.
2 Lambers TT, Gloerich J, van HE, Alkema W, Hondmann DH, van Tol EA: Clustering analyses in peptidomics revealed that peptide profiles of infant formulae are descriptive. Food Sci Nutr 2015;3:81-90.
3 von Berg A, Koletzko S, Grubl A, Filipiak-Pittroff B, Wichmann HE, Bauer CP, Reinhardt D, Berdel D: The effect of hydrolyzed cow's milk formula for allergy prevention in the first year of life: the German Infant Nutritional Intervention Study, a randomized double-blind trial. J Allergy Clin Immunol 2003;111:533-540.
4 von Berg A, Koletzko S, Filipiak-Pittroff B, Laubereau B, Gruebl A, Wichmann HE, Bauer CP, Reinhardt D, Berdel D: Certain hydrolyzed formulas reduce the incidence of atopic dermatitis, but not of asthma: Three year results of the GINI-Study. J Allergy Clin Immunol 2007;119:718-722.
5 von Berg A, Filipiak-Pittroff B, Kramer U, Link E, Bollrath C, Brockow I, Koletzko S, Gruebl A, Heinrich J, Wichmann HE, Bauer C, Reinhardt D, Berdel D, and the GINIplus study group: Preventive effect of hydrolyzed infant formulas persists until age 6: long-term results from the German Infant Nutritional Intervention Study GINI. J Allergy Clin Immunol 2008;121:1442-1447.
6 von Berg A, Filipiak-Pittroff B, Kramer U, Hoffmann B, Link E, Beckmann C, Hoffmann U, Reinhardt D, Gruebl A, Heinrich J, Wichmann HE, Bauer CP, Koletzko S, Berdel D: Allergies in high-risk schoolchildren after early intervention with cow milk protein hydrolyzates. 10-years results from the German Infant Nutritional Intervention Study GINI. J Allergy Clin Immunol 2013;131:1565-1573
7 von Berg A, Filipiak-Pittroff B, Schulz H, Hoffmann U, Link E, Sussmann M, Schnappinger M, Bruske I, Standl M, Kramer U, Hoffmann B, Heinrich J, Bauer CP, Koletzko S, Berdel D: Allergic manifestation 15 years after early intervention with hydrolyzed formulas - the GINI Study. Allergy 2016;71:210-219.
8 Lowe AJ, Hosking CS, Bennett CM, Allen KJ, Axelrad C, Carlin JB, Abramson MJ, Dharmage SC, Hill DJ: Effect of a partially hydrolyzed whey infant formula at weaning on risk of allergic disease in high-risk children: A randomized controlled trial. J Allergy Clin Immunol 2011;128:360-365.
9 von Berg A, Kramer U, Link E, Bollrath C, Heinrich J, Brockow I, Koletzko S, Grubl A, Filipiak-Pittroff B, Wichmann HE, Bauer CP, Reinhardt D, Berdel D: Impact of early feeding on childhood eczema: development after nutritional intervention compared with the natural course - the GINIplus study up to the age of 6 years. Clin Exp Allergy 2010;40:627-636
10 Boyle RJ, Tang ML, Chiang WC, Chua MC, Ismail I, Nauta A, Hourihane J, Smith P, Gold M, Ziegler J, Peake J, Quinn P, Rao R, Brown N, Rijnierse A, Garssen J, Warner. Probiotic-supplemented partially hydrolysed cow's milk formula for the prevention of eczema in high risk infants: a randomised controlled trial. J Allergy Clin Immunol 2016 (in press)
Sibylle Koletzko, MD, PhD1
Birgit Filipiak-Pittroff, MSc2
Joachim Heinrich, PhD2
Berthold Koletzko, MD, PhD1
Dietrich Berdel, MD, PhD3
Andrea von Berg, MD3
1 Dr. von Hauner Children’s Hospital, University of Munich Medical Centre, Ludwig Maximilians University, Munich, Germany; 2 Institute of Epidemiology I, Helmholtz Zentrum Munich, 3Marien-Hospital Wesel, Department of Paediatrics, Wesel, Germany
Corresponding author:
Prof. Dr. Sibylle Koletzko
Div. Gastroenterology and Hepatology
Dr. v. Hauner Children’s Hospital,
Ludwig Maximilians University Munich
Lindwurmstr. 4
D-80337 Munich, Germany
e-mail: sibylle.koletzko@med.uni-muenchen.de
Phone +49 89 44005 2811
Fax +49 89 44005 7898
E-mail: sibylle.koletzko@med.uni-muenchen.de.
Competing interests: The GINI Intervention study was funded for 3 years by grants from the Federal Ministry for Education, Science, Research and Technology (grant no. 01 EE 9401-4). The companies Milupa, Nestlé, Mead Johnson and Nutricia provided the blinded study formulas for the participating children for the first 4-6 months. The 3-6 and 10-year follow-up examinations of the GINI study were covered from the respective budgets of the initial 4 study centres (Wesel, LMU Munich, TU Munich, Helmholtz Center Munich (former GSF)). From 6 years onwards, GINI was additionally partly supported by the Federal Ministry for the Environment (IUF, FKZ 20462296) and by the budget of the IUF- Leibniz Research Institute for Environmental Medicine at the University of Düsseldorf. The 15-year follow-up of the GINI study was supported by the companies Mead Johnson and Nestlé and by cooperation in European Studies (e.g MeDALL, ESCAPE). Some projects not directly related to the intervention effect of the hydrolyzates (e.g. effect of cesarean section, effect of solid food introduction) were partly supported by Nestlé, Mead Johnson, Numico, Pharmacia and Stiftung Kindergesundheit, and in cooperation with European studies (e.g MeDALL).
Pardon my ignorance if I missed this in your study.
Were the kids given hydrolyzed formula for the first year of life then cows milk at a year of age? Also, at what point in the study were kids under a year exposed to dairy products?
Competing interests: No competing interests
Re: Hydrolysed formula and risk of allergic or autoimmune disease: systematic review and meta-analysis
We thank Drs Sibylle Koletzko, Filipiak-Pittroff, Heinrich, Berthold Koletzko, Berdel and von Berg for their comments on our article, and are pleased to have the opportunity to clarify some aspects of our work.
Koletzko et al first raise the issue whether meta-analysis of different brands of hydrolysed formula milk is appropriate. While we did stratify our analyses according to degree of hydrolysis (partial versus extensive) and protein source (casein versus whey), we did not make an a priori assumption that a specific formula such as partially hydrolysed whey formula (pHF-whey) from one manufacturer would have different health effects to pHF-whey from a different manufacturer. For the eczema outcome which the authors helpfully summarise in their table, we did undertake a post hoc analysis of the Nestlé pHF-whey, since this formula was used in 7 of the 11 randomised controlled trials addressing whether pHF in place of standard formula influences risk of eczema in the first 4 years of life. The attached Figure shows this analysis, where we found no evidence that this specific pHF-whey influences risk of eczema, and no statistical heterogeneity between study findings including those of the Lowe et al trial discussed below.
The second issue raised by Koletzko et al is whether the study of Lowe et al should have been classified as high or unclear risk of assessment bias. We classified this study as at low risk of assessment bias, since Lowe et al state in their publication ‘Staff were blind to these allocation codes and to the group of allocation at the time of outcome assessment’. Thus outcome assessment was described in the publication as blind to treatment allocation. The study of Lowe et al was classified as at unclear risk of bias overall, because the procedure for allocation concealment was not clearly described, and was classified as at high risk of conflict of interest because it was funded by the formula manufacturer Nestlé.
Koletzko et al helpfully highlight the risk of attrition bias in their own study, due to funding restrictions leading to exclusion of some participants from outcome assessment. We classified attrition bias as low in their study, despite significant loss to follow up at 3 years, due to the use of Generalised Estimating Equation for data analysis, which includes data from earlier timepoints for those participants with missing 3 year outcome data. We classified their study as at unclear risk of conflict of interest, because several authors including Drs von Berg and Koletzko report receiving speaker fees and research support for themselves or their institutions from manufacturers of hydrolysed infant formula, and Dr Koletzko serves on the advisory board for Danone, Nestlé and Mead-Johnson. The study was not classified as high risk of conflict of interest, because the initial study was independently funded.
Formula manufacturer, risk of bias and conflict of interest are just three sources of potential heterogeneity between studies. Where possible we explored our findings for several sources of heterogeneity in this report, and took those into account in drawing our conclusions. Our key subgroup analyses were type of hydrolysed formula, risk of bias, risk of conflict of interest, study design, type of intervention, and population disease risk.
The final issue raised by Koletzko et al is our selection of data from their trial. They question the use of intention to treat data from their study rather than data that exclude participants who exclusively breastfed to at least 4 months, or were non-compliant with the study protocol in other ways (per protocol analysis). The authors cite their exclusive breastfeeding rate at 4 months of 39.5% as high, but this is similar to the global average of 37% in the first 6 months of life (1). In a study where randomisation occurred at birth, any exclusion of participants following birth will mean that the protection from bias afforded by randomisation is lost. Thus intention to treat analysis is the least biased approach to data interpretation. As Koletzko et al point out, the characteristics of mothers and infants who use infant formula are different from the characteristics of mothers and infants who don’t use infant formula in some studies – so exclusion of participants based on their level of infant formula use may introduce bias into the analyses. Koletzko et al also question our use of clinically assessed eczema versus parent reported eczema from their trial. Substitution of one for the other does not materially change the findings of our meta-analysis, with no statistically significant reduction in eczema at age 0-4 years in either case.
While we acknowledge that the studies included in this systematic review do suffer from significant methodological issues and risk of conflict of interest, our report currently represents the most rigorous analysis available in this area. Our findings do not support the current widespread recommendations to use hydrolysed formula in infants at high risk of allergic conditions.
1. Victora CG, Bahl R, Barros AJ, Franca GV, Horton S, Krasevec J, et al. Breastfeeding in the 21st century: epidemiology, mechanisms, and lifelong effect. Lancet. 2016;387(10017):475-90.
Competing interests: No competing interests