Discussion
The clinical phenotype of FPIES is well described, with several large-scale retrospective cohort studies published over the last two decades. Despite this, the reason why there are prominent geographic variations in types of food triggers, rates of multiple food FPIES and why children outgrow FPIES remain unknown. Regional breastfeeding and weaning practices have been proposed to influence these factors.
We have previously investigated the impact of infant feeding practices on the development of FPIES to multiple triggers, in addition to the associations among different food groups in infants with multiple food FPIES 2. Rice was the most common food trigger in Australian children, with 32% of the total cohort having FPIES to multiple food triggers. Here, we demonstrate that rice is consistently the most common trigger in Australian infants (in 45%). This is in contrast to other regions with cow’s milk being the most common in the United Kingdom, USA and Israel, and with fish being a predominant solid food trigger in Italy and Spain 3, 5, 6, 8, 12. Only one other study outside of Australia, in the USA, identified rice as more frequent a trigger than cow’s milk 13. Interestingly, in our cohort, infants with rice FPIES were unlikely to react to cow’s milk or fish but were more likely to have FPIES to multiple foods.
We identified a rate of FPIES to multiple food triggers within the whole cohort of 36% in keeping with previous Australian data2, and in comparison with other countries, highest in the USA (up to 50%) and United Kingdom (30%) 5, 14, and lowest in Spain (16%) and Italy (15%) 6, 15. Here we demonstrate a possible trend towards increased risk of FPIES to multiple triggers in children who were delivered via caesarean section. The mode of delivery has long been proposed to impact upon the development of food allergy and atopy 16, with delivery via caesarean section thought to disrupt the maternal transference of microbiota to the infant. Katz et al. reported a higher prevalence of delivery via caesarean section in infants with cow’s milk FPIES, two times compared to vaginal delivery 3.
Our previous population-based study demonstrated that duration of exclusive breastfeeding less than four months was associated with increased risk of multiple food FPIES (11% versus 20%)2. This was not seen within this study’s cohort. Instead, there was a weak association between increased duration of breastfeeding and increased risk of multiple food FPIES. It may be that mothers of children who have experienced multiple FPIES may choose to breastfeed for longer duration, due to the perceived relative safety of breastfeeding in triggering an FPIES reaction, thus a potential for reverse causation. Few studies have examined infant feeding practices in association with FPIES, and those which have, demonstrate similar or mildly higher rates of FPIES in infants who are not breastfed, or have shorter duration of breastfeeding 5, 6. Overall the rate of breastfeeding in this population is high, compared to other countries with well documented FPIES cohorts. Feeding practices and their impact on FPIES development and phenotype warrant further prospective investigation.
Co-associations between food triggers in children with multiple FPIES were observed between rice and oats; cow’s milk and soy; and fish and shellfish, consistent with previous data in reported case series17. We observed a co-association between eggs and other grains (which included wheat and rye) in our cohort not previously documented in other reports. Cow’s milk/soy cross-reactivity is commonly seen in US populations (up to 50% in case series), as is rice/oats cross-reactivity (up to 30%) in US and Australia 1. Although fish and shellfish FPIES are more prevalent in Mediterranean countries, the reported rates of cross-reactivity are similar18. Children with rice and cow’s milk FPIES had an increased risk of FPIES to multiple foods. This may be due to rice and cow’s milk being the most common food triggers within the cohort, in addition to the higher prevalence of co-association between rice/oats and cow’s milk/soy.
Atopy in siblings was another risk factor for multiple FPIES with a 1.6-fold risk. In contrast, a family history of FPIES was rare and not associated with increased risk of multiple FPIES, although the numbers for this analysis were small. This is similar to previous epidemiological data demonstrating low familial risk of FPIES19. Both a family and personal history of atopy is common in Australian infants and although not examined in local populations, a US study has reported a higher rate of comorbid atopy in infants with FPIES compared to healthy infants 20. In our cohort, eczema was the most common comorbid atopy and was strongly associated with IgE-mediated food allergy, particularly to eggs, similarly demonstrated by Ruffner et al. 20. Additionally, there is a risk of conversion from FPIES to IgE food allergy, especially with cow’s milk FPIES 1. It would be of interest to conduct a longitudinal study to observe for the rates of development of IgE food allergy, given the apparent existing high background risk.
Proposed definitions for severity of FPIES reactions have been characterised by international guidelines 1, and few descriptive case series have observed potential increased severity in association with solid food triggers, however reports are varying. In our cohort, the rate of experiencing at least one severe reaction was high, at nearly 75% of the cohort, however no risk factors for severity were identified.
We have previously reported variations in timing to acquisition of tolerance to different causative food triggers 21. Fifty five of 169 infants were participants in this previous study21. Here, we demonstrate similar findings of earlier resolution for rice FPIES but delayed resolution for fish FPIES. Infants who had their first reaction at a later age attained tolerance at a later age, likely related to the sequence of exposure to different solid foods. Although we did not find any predictive factors of tolerance from feeding history, Sopo et al. demonstrated that earlier tolerance acquisition in FPIES to eggs was achieved by the effect of cooking22, providing potential insight into how food protein interactions with immune mechanisms may dictate responses. Despite a proof of tolerance via supervised oral food challenge (OFC), a recent study by Argiz et al. reported reactions upon re-exposure to trigger foods following a negative OFC in 13% of infants, most likely to occur with egg and fish FPIES 23. These findings aid in recommendations for feeding practices and first weaning foods for infants with FPIES.
Our study is limited by the retrospective nature, which is prone to recall bias. In addition, our sample size is limited by selection of a group of individuals known to a tertiary specialised centre, with possible identification of a more severe cohort of patients with FPIES, as illustrated by the high proportion of subjects who experienced at least one severe episode. Missing data on mode of birth, and incomplete feeding histories limit the conclusions which can be drawn about these analyses.
Our findings over 20 years of children with FPIES managed at a large single centre demonstrate many similarities with the published population-based survey of Australian infants with FPIES, which examined only FPIES presentations from 2012-14. Notably, we observed decreased odds of the development of multiple food FPIES and severe reactions in infants who were born via vaginal delivery, and a possible variable effect of breastfeeding, with longer duration of breastfeeding seen in infants with multiple food FPIES, but with a decreased risk of development of severe reactions. Important clinical implications include identifying that infants presenting with cow’s milk or rice FPIES may be at greatest risk of the development of FPIES to other foods and that severity of FPIES reactions appears not to be readily predictable by food trigger or other baseline characteristics.