Prevention

The development of food allergy depends on several factors, including genetic factors and early exposure to allergenic proteins in the diet, food protein uptake and handling, and the development of tolerance. Many hypotheses, as regards the possible causal relationships, have been raised during the past few years, including the hygiene theory, the role of bacterial gut flora, and the potential effect of different cytokines in breast milk. Although interesting, these are mainly speculations based on non-interventional and often retrospective/cross-sectional studies including small study populations. These theories remain to be documented in proper, controlled and prospective studies. Breastfeeding and the late introduction of solid foods (>4 months) is associated with a reduced risk of food allergy, atopic dermatitis, and recurrent wheezing and asthma in early childhood. In all infants, breastfeeding should be encouraged for 4-6 months. In high-risk infants a documented extensively hydrolysed formula is recommended if exclusive breastfeeding is not possible for the first 4 months of life. There is no evidence for preventive dietary intervention neither during pregnancy nor lactation. Preventive dietary restrictions after the age of 4-6 months are not scientifically documented.


Introduction
The development and phenotypic expression of atopic disease depends on an interaction between genetic factors, environmental exposure to food and inhalant allergens, and non-speci®c adjuvant factors (e.g. tobacco smoke, air pollution and infections). The expression of allergic diseases may vary with age, and symptoms may disappear and be replaced by other symptoms. In infancy, the main atopic symptoms are atopic dermatitis, gastrointestinal symptoms, and recurrent wheezing, whereas bronchial asthma and allergic rhinoconjunctivitis are the main problems later in childhood. Adverse reactions to foods, mainly cow's milk protein, are most common in the ®rst years of life, whereas allergy to inhalant allergens mostly occurs later. It is thus very important to take into consideration the natural course of the disease when evaluating the results from interventional studies on the prevention of allergic diseases.
The development of food allergy/food intolerance (FA/ FI) depends on several factors, including early exposure to allergenic proteins (time, dose, frequency), food protein uptake and handling, and the development of tolerance. The development and rather high incidence of FA/FI and especially cow's milk protein allergy/ intolerance (CMPA/I) in infancy has been suggested to be caused by an incomplete mucosal barrier, increased gut permeability to large molecules and the immaturity of local and systemic immunological responses. Human colostrum/milk facilitates maturation of the gut and provides a passive protection against bacteria and antigens (allergens) by means of speci®c secretory IgA and other protecting factors [1].
When evaluating possible risk factors for the development of allergic diseases and the possible effect of preventive measures, it is important to stress some factors. The study design of both non-interventional and interventional studies should be prospective (true), including well-de®ned diagnostic criteria, a suf®cient duration of follow-up, and a proper sample size for adequate statistical evaluation. Besides, interventional studies should include proper randomization, double blinding, and a control for confounders, as well as the proper registration of compliance and follow-up of dropouts.
As a result of recall bias and selection bias, retrospective studies should not be used for evaluation of predictive/ risk factors for the development of allergic diseases. Likewise, cross-sectional studies are not suitable for the assessment of cause±effect relationships between exposure to allergens/adjuvant factors and the development of allergic diseases. Generally, it should also be borne in mind that prospective non-interventional studies can be used to generate hypotheses on the relationship between cause and effect in the development of allergic diseases. However, proper con®rmation of a possible cause±effect relationship requires a demonstration of the causative mechanism and effect of the elimination/prevention of the suspected causative factor.

Non-interventional studies Atopic heredity
From prospective studies many possible predictive factors of the development of allergic diseases have been identi®ed. Although it is well documented that atopic heredity is associated with an increased risk of the development of allergic diseases [4±7], it has also been demonstrated that most children who develop atopic symptoms during the ®rst years of life come from families without an atopic heredity. The majority of children with recurrent wheezing/asthma thus do not belong to high-risk groups for the development of atopic disease [4], whereas a higher proportion of children with allergic disease, including other manifestations than asthma, will have atopic heredity [7].

Dietary factors Breastfeeding
Previous prospective non-interventional studies in unselected infants have shown an association between early infant feeding of cow's milk formula and the development of atopic eczema and CMPA/I (Table 1) [8±10,11 . . ]. One cohort study [8] demonstrated an association between early cow's milk formula feeding and the development of cow's milk protein allergy. A previous non-interventional study [9] indicated a preventive effect of breastfeeding as regards eczema and food allergy at 1 and 3 years and asthma up to 17 years of age, but no adjustment for confounders were performed. A recent prospective Australian study [10] including 2979 children showed a substantial reduction in the risk of childhood asthma assessed at 6 years of age, if exclusive breastfeeding was continued for at least 4 months of life. In that study, logistic regression analysis including important confounders was used. However, it is important to bear in mind that several studies have shown that mothers who choose to breastfeed their infants behave in another way compared with mothers who choose formula feeding. Exclusively breastfed infants also belong to a higher social class, are less exposed to tobacco smoke and pets, and have solid foods introduced at a later age, which are factors shown to affect the risk of the development of atopic disease, especially respiratory symptoms [12 . . ].
A possible protective effect of breastfeeding on the development of atopic diseases may be caused either by: (i) a protective effect of human milk (the constituents) or (ii) the avoidance of a`high dose' of cow's milk proteins.
Low concentrations of food allergens especially cow's milk proteins are shown to be present in human milk when consecutive testing is done. According to the low incidence of CMPA/I in exclusively breastfed infants at 0.5% in unselected infants [8], and 1.3% in high-risk infants [12 . . ] in prospective birth cohort studies, this low-degree exposure seems to induce tolerance rather than disease.
Recent studies indicated that variations in the composition of human milk, e.g. low levels of alpha-linolenic acid and a disturbed relationship between the n-3 and the n-6 fatty acids [13] or varying concentrations of cytokines [14 . ,15 . ], may partly explain some of the controversies regarding the protective effect of breastfeeding against allergy. These observations are based on the analyses of samples of breast milk from mothers on a restrictive diet because of diagnosed CMPA/I in their children compared with a small number of samples from mothers on an unrestricted diet and children without CMPA/I. No consecutive testing of breast milk was performed. No association between nucleotide and polyamine levels in human milk and atopy development during the ®rst year of life has been found [16]. These observations are interesting, but not conclusive, and prospective long-term follow-up studies with a proper sample size are desirable for con®rmation of these possible relationships.

Solid foods
As for the introduction of cow's milk proteins before 4 months of age, the introduction of complementary foods (solid foods) before 4 months of age has been associated with a higher risk of atopic dermatitis up to the age of 10 years [17].

Intestinal microbial flora
It has been hypothesized that the intestinal microbial¯ora may in¯uence the development of sensitization [18,19 . ,20 . ]. Two of the studies [18,19 . ] are crosssectional, including a small number of children at the ages of 2 years and 13 months with and without atopic symptoms. In one prospective study of 76 atopic predisposed infants [20 . ], a signi®cant difference in the bacterial cellular fatty acid pro®le of stool samples was found by 3 weeks, but not at 3 months of age in atopic infants compared with non-atopic infants. Atopic infants (n = 18) were de®ned as having at least one positive skin-prick test to any of 16 allergens at the age of 12 months. The authors concluded that differences in the neonatal gut micro¯ora precede the development of atopy, suggesting a crucial role for the balance of indigenous intestinal bacteria for the maturation of human immunity to a non-atopic mode. Meanwhile, the study did not provide ®rm evidence for such a conclusion. Convincing evidence for this theory, as well as the hypothesis (speculation) that an unclean lifestyle with lots of infections during early childhood may prevent the development of atopic disease, needs con®rmation by prospective studies.

Interventional studies Dietary measures Unselected/non-high-risk infants
Only a few prospective intervention studies have been performed in infants without a hereditary atopic predisposition ( was not randomized, and the outcome was based on data from medical ®les and questionnaires obtained at 7, 11 and 14 years. In one study [24] the diagnostic criteria were unspeci®c, and outcome measures were mainly based on questionnaires. In a recent randomized study including a large number (n = 6209) of full-term, unselected newborns [25,26 . ], the infants with a need for supplementary formula received either adapted liquid cow's milk formula, pasteurized human milk or extensively hydrolysed formulas (eHF). The results indicated that the feeding of cow's milk-based formula at the maternity hospital increased the risk of CMPA when compared with feeding an eHF, but exclusively breastfeeding for 8 weeks did not eliminate the risk of CMPA. Besides, the results indicated that signi®cant risk factors for the presence of IgE antibodies to cow's milk in allergic infants were a long duration of breastfeeding, exposure to cow's milk at the maternity hospital, and breastfeeding during the ®rst 2 months at home either exclusively or combined with infrequent exposure to small amounts of cow's milk [26 . ]. However, the dietary intervention only included the ®rst 4 days of life, after which period the mothers chose the diet themselves, and it is hard to exclude the possibility that this`self-selection' in¯uenced the results.
Another prospective study to investigate the overall health bene®ts of an allergen-reduced dietary regimen in a large unselected study population has so far resulted in two publications during the past year [27 . ,28 . ]. That open study included 1130 unselected healthy newborns allocated (not randomly assigned) to an intervention and a non-intervention cohort according to the place of birth. In the intervention group, breastfeeding or feeding moderate whey hydrolysate formula with no weaning food before 4 months of age was recommended, whereas the control group had no intervention. The authors concluded that an allergen-reduced dietary recommendation that includes a moderate whey hydrolysate infant formula has no negative effects on infant growth and leads to improved general health status, mainly because of improvements in skin ®ndings when compared with a ]. However, the two study cohorts represented two different regions and were signi®cantly different as regards parent education, household pets, the number of older siblings, exposure to tobacco smoke and urban residence. The study allows no conclusions as regards allergy prevention. It is astonishing that such an intervention study including an allergen-reduced diet did not include speci®c allergic disease as an outcome measure.

High-risk infants
An effect of dietary allergy prevention has only been demonstrated in high-risk infants, i.e. infants with double parental atopy or at least one ®rst-degree relative (parent or sibling) with documented atopic disease (doctor diagnosed), the latter possibly combined with elevated cord blood IgE in the case of single atopic predisposition. It is dif®cult to compare the effect of these different prevention programmes because of the great variations in study design and diagnostic criteria [1,29].

Breastfeeding
In high-risk infants, exclusively breastfeeding for 4 months in combination with the avoidance of solid foods has resulted in a signi®cant reduction of the cumulative incidence of CMPA/I and atopic dermatitis during the ®rst 4 years of life [21,30±35] (  (Table 4).

Formulas
Results from prospective studies including formula feeding are summarized in Table 5. Some prospective studies have shown that soy formulas are as allergenic as conventional cow's milk-based formulas, and on this basis they should not be recommended for the prevention of food allergy [29,33], but different views exist [ 45± 47]. There is no evidence that formulas based on whole proteins other than cow's milk protein are less allergenic.
Several prospective studies [12 . . ,33,35,43,44,48±54] demonstrated a preventive effect of eHF in combination with the avoidance of cow's milk proteins and solid foods during at least 4 months in high-risk infants on the cumulative incidence of atopic dermatitis and food allergy, especially CMPA/I until the age of 4±7 years. A real preventive effect, and not only a postponement of the onset of the disease, was thus documented.
In randomized prospective studies [32,34,53±57] in highrisk infants an allergy preventive effect of partly hydrolysed formulas (pHF) has been reported. Because of great variations in study design and diagnostic criteria, the relative ef®cacy of the different interventions tested in the various studies cannot be compared directly with each other, and only a few studies comparing the effect of eHF and pHF have been published. A recent Swedish study [58] reported a lower cumulative incidence of atopic symptoms up to the age of 18 months with both an eHF and a pHF diet, compared with a cow's milk protein-based formula; a greater effect was reported with eHF. These data are supported by a Danish prospective randomized, double-blind study Effect of infants' diet indicated by arrows: : indicates increase, ; indicates decrease or ? no change in the incidence of disease/symptoms in infants having the diet mentioned. a Premature infants; b Low birthweight infants; c Included diet only for the first few days. CMPA, Cow's milk allergy; CMF, cow's milk-based formula; eHF, extensively hydrolysed formula; HM, human milk; RAST, radioallergosorbent test. from birth to 18 months of age comparing the allergy preventive effects of two eHF and one pHF, published last year [12 . . ]. The study included a 1-year birth cohort of high-risk infants (n = 478) randomly assigned at birth to one of the three hydrolysed formulas if breastfeeding was not possible or suf®cient. Only a few infants were never breastfed, most of the infants were breastfed exclusively (n = 232) or had varying amounts of supplement with formula (n = 246) until the age of 4 months. The introduction of complementary foods was not recommended until the age of 4 months. After the age of 4 months no dietary restrictions were recommended, and all mothers had an unrestricted diet during pregnancy and lactation. In the study the overall incidence of con®rmed CMPA/I was low (1.3%) and signi®cantly lower in infants fed eHF (0.6%) compared with pHF (4.7%). The study thus showed a better allergen-speci®c preventive effect of eHF, but did not exclude an effect of a pHF also.   months of life only, and this regimen resulted in a very low incidence of CMPA/I and other food allergies until the age of 18 months. In one recent study [59], no signi®cant effect of supplement with eHF compared with cow's milk-based formula after the age of 6 months in breastfed high-risk infants was found. Controlled studies concerning the possible preventive effect of the avoidance of other potential food allergens, e.g. egg, ®sh, etc., after the age of 4±6 months of life have not been published. There is thus no evidence of an allergy preventing effect of restrictive diets after 6 months of age.

Conclusion
As summarized in Table 6, recent prospective interventional and non-interventional studies have demonstrated a preventive effect of simple dietary measures during the ®rst 4±6 months of life as regards the development of food allergy, especially cow's milk allergy and atopic eczema.

References and recommended reading References and recommended reading
Papers of particular interest, published within the annual period of review, have been highlighted as: . of special interest    . A double-blind randomized prospective intervention study of a 1 year birth cohort of high-risk infants. The intervention included feeding breastfeeding or one of three blinded hydrolysed formulas ± one pHF and two eHF ± according to randomization at birth. A low incidence of CMPA/I and a significantly better effect of eHF compared with pHF was found.