Introduction

Non-IgE-mediated food allergy encompasses a wide range of disorders affecting the gastrointestinal tract (food protein-induced enterocolitis (FPIES) syndrome, allergic proctocolitis, enteropathy, celiac disease, cow’s :milk allergy-induced iron-deficiency anemia); skin (contact dermatitis to foods, dermatitis herpetiformis); and lungs (Heiner’s syndrome, also known as pulmonary hemosiderosis). In addition, non-IgE-mediated food allergy has been implicated as a potential culprit in a subset of patients with gastroesophageal reflux disease, chronic emesis and diarrhea, constipation, and irritable bowel syndrome [1]. This review will discuss three non-IgE-mediated disorders affecting the gastrointestinal tract: FPIES, food protein-induced allergic proctocolitis (FPIAP), and food protein-induced enteropathy (FPE).

Definitions and Classification

Consensus statements from experts in food allergy delineated the defining features of these disorders [2]. FPIES is typically characterized by profuse emesis, diarrhea, dehydration, and lethargy. It may be either insidious in onset with chronic exposure to the inciting food, or more acute, within onset of 1–3 h after ingestion if exposure is intermittent [3, 4]. FPIAP is characterized by blood-streaked stools in otherwise healthy, thriving infants [5, 6]. FPE also manifests in infancy, with chronic diarrhea, vomiting, malabsorption, and hypoproteinemia [7, 8]. Features of these disorders are compared in Table 1.

Table 1 Features of food protein-induced enterocolitis (FPIES), allergic proctocolitis (FPIAP), and enteropathy (FPE)
Full size table

Epidemiology

The prevalence and epidemiological features of these disorders have not been well studied. Available prevalence data suggest that these are less common than IgE-mediated food allergies.

While FPIES has been considered a somewhat rare diagnosis, available data suggest that FPIES may be more prevalent than previously assumed. Among an Israeli birth cohort, milk-FPIES was reported in 0.34 % of 13,019 infants; by comparison, IgE-mediated food allergy to milk in this population was 0.5 % [9]. Preliminary data from an Australian registry currently estimates prevalence at 1 in 10,000 children less than 2 years of age [10]. However, this may be an underestimate as reporting is voluntary and accurate reporting could be hampered by incomplete understanding of FPIES among general practitioners [11]. There is some suggestion that prevalence may be increasing, though it is unclear whether this is related to improved recognition or a true increase in prevalence [12–14].

FPIAP similarly is without sufficient prevalence studies, but is thought to be a common cause of rectal bleeding in infancy. In a prospective study from Ohio, which included 22 otherwise healthy infants with rectal bleeding, 64 % were found to have evidence of FPIAP on endoscopy with biopsies [15]. However, infants were not challenged to confirm the diagnosis. The best data regarding prevalence of FPIAP come from the prospective population-based study from Israel, which reported cow’s milk FPIAP prevalence of 0.16 % among 13,019 infants [16•]. When authors limited diagnosis to those infants whose diagnosis was confirmed with food challenge, prevalence was much lower. Of 14 infants who were challenged with cow’s milk at a median of 3 months following initial presentation, only 3 developed reoccurrence of rectal bleeding. Lack of symptom reoccurrence in the majority of infants speaks either to the transience of sensitivity to milk in FPIAP or to the possibility that milk sensitivity did not account for rectal bleeding in these infants. Significant discrepancies between diagnosis of presumed proctocolitis and challenge-proven proctocolitis are discussed further under a subsequent subheading.

The prevalence of FPE is even more obscure. Investigators have noted a decline in the prevalence of this entity over the past few decades [17•]. This may be due to true decrease in prevalence as exclusive breastfeeding has become more common in the last few decades, or possible underreporting with prompt formula changes for affected infants before a formal diagnosis is made.

Clinical Manifestations and Presentation

FPIES, FPIAP, and FPE are typically disorders of infancy that manifest with gastrointestinal symptoms, and notably lack cutaneous and respiratory findings seen in IgE-mediated food allergies. Symptoms resolve with removal of the offending protein from the diet (or maternal diet in exclusively breastfed infants). The most common triggers are cow’s milk and soy, though a variety of other foods may also be triggers.

FPIES usually presents in infancy and childhood with frequent emesis, and may be either acute or chronic. In the chronic form, which is more common in formula-fed young infants with frequent exposure to the inciting protein, intermittent emesis, diarrhea, failure to thrive, and dehydration develops. The acute form, which occurs with intermittent exposure, is more severe in presentation and is characterized by profuse emesis, dehydration, and lethargy with onset of 1 to 3 h after ingestion of the culprit food [3, 18]. Previous studies established that presentation of FPIES may be quite severe; in addition to profuse emesis and dehydration, infants may develop lethargy, cyanosis, hypotension, hypothermia, and methemoglobinemia [19–23, 24•]. However, these severe symptoms continue to be under-recognized as possible manifestations of FPIES [11, 25]. Studies published in the last several years from referral centers confirm that severe manifestations of FPIES reflecting hypotension and poor oxygen delivery are fairly common, with lethargy, hypotonia, pallor, or cyanosis noted in 40 to 85 % of patients at presentation [9, 26–28].

Though FPIES has been thought of as a pediatric disease, recent publications suggest that FPIES may also develop in adulthood. In 2012, Fernandes et al. reported a case of a 53 year old with FPIES to scallop, which was confirmed with positive oral food challenge and supported by blood work during the challenge showing leukocytosis and neutrophilia with a normal tryptase [29]. In a recent case series published in 2014, Tan et al. report 31 cases of acute onset of vomiting, abdominal pain, and diarrhea triggered by a specific food in patients over 18 years of age [30]. Mean age of symptom onset was 29 years. Symptoms started 1 to 2 h after eating the food, lasted 1 to 3 h, and occurred on a median of two occasions with the suspected food [30].

FPIAP typically presents in infancy with blood-streaked and mucus stools [31, 32]. Frank diarrhea is less common, and was reported in 28.3 % of infants diagnosed with FPIAP in a recent meta-analysis [32]. In contrast to FPIES and FPE, infants with FPIAP are generally well appearing and thriving, without systemic symptoms [5]. Additionally, many infants with FPIAP are breastfed, approximately 50 % from a recent meta-analysis [32]. Infants generally experience improvement in symptoms within 3 days of removal of the offending protein from the maternal diet or change to the hypoallergenic formula, though complete resolution may take up to 2 weeks [33]. Symptoms generally develop in the first 2 to 8 weeks of life, but can develop in the first few days [6]. Though FPIAP has generally been thought of as a disease of infancy, cases have been described in older children. Ravelli et al. reported 16 cases of rectal bleeding that resolved with elimination of cow’s milk protein in children aged 2 to 14 years. Endoscopic and histologic findings were additionally consistent with FPIAP [34]. It is not clear, however, whether this is the same clinical entity. In young adults, eosinophilic colitis (EC) has a more severe and chronic relapsing course and is rarely associated with food allergy. Additionally, endoscopic findings are distinct from FPIAP; an intense eosinophilic infiltration in the colon can be segmental or diffuse and may affect several intestinal layers [35].

FPE also presents in infancy, with the most prominent symptoms being diarrhea and failure to thrive, often accompanied by emesis and abdominal distension [7, 8]. Distinguishing features include malabsorption with steatorrhea (in up to 80 % of infants) [17•] and lack of acute symptoms as seen in cow’s milk-FPIES.

Diagnosis

Current diagnostic criteria from the NIAID-sponsored panel in 2010 advise that diagnosis for FPIES, FPIAP, and FPE be based on typical presenting features, resolution with removal of the offending protein, and reoccurrence of symptoms with food challenge (OFC). OFC with the offending protein may not be necessary, however, when history is convincing or the reaction was severe (as in acute FPIES) [18]. An International Expert Panel Work group of Academy of Allergy, Asthma, and Immunology is expected to update consensus guidelines for diagnosis as well as management, with projected completion in the coming year.

Diagnosis of these three entities can be challenging, as all three are without confirmatory diagnostic tests. On initial presentation when symptoms are ongoing and particularly, if the infant has systemic symptoms, it is important to consider and exclude other entities in the differential diagnosis. Skin tests and serum food-specific IgE are not useful in establishing the diagnosis and are negative in a majority of patients at presentation [18, 27, 36]. A notable exception to this is FPIES with concomitant IgE sensitization, which is discussed further below.

In FPIES, the ill appearance of the infants on initial presentation warrants basic studies including complete blood count, blood gas, chemistries, and co-oximetry as part of a diagnostic work-up. For infants presenting with vomiting and lethargy, these studies may help in narrowing the broad differential diagnosis, which includes sepsis, infectious gastroenteritis, necrotizing enterocolitis, bowel obstruction, inborn errors of metabolism, celiac disease, inflammatory bowel disease, eosinophilic gastrointestinal disease, and anaphylaxis [37]. Laboratory findings consistent with acute or ongoing FPIES reaction include leukocytosis with neutrophilia, thrombocytosis, non-anion gap acidosis, methemoglobinemia, and hypoalbuminemia [3, 13, 22]. Stool studies in symptomatic infants may reveal leukocytes, frank or occult blood, and eosinophils. A recent case series investigated use of fecal eosinophil-derived neurotoxin as a biomarker, and found that levels of this neurotoxin were significantly elevated 12 h after exposure to inciting food, with a peak at 24 h. Limitations of this study include small sample size and lack of a control group, making it unclear whether this test would help distinguish FPIES from other causes of clinical enterocolitis [38].

While the laboratory studies above can be helpful in the setting of an ongoing reaction, laboratory work-up is generally normal after a reaction has resolved. As yet, we are without diagnostic testing other than food challenge that would confirm or suggest a diagnosis of FPIES once an acute reaction has resolved. Atopy patch testing (APT) was suggested as a promising predictor of positive food challenge in a small study [39•, 40]; however, subsequent larger studies did not find APT to have any predictive value [24•, 41]. More recently, Konstantinou et al. found deficient T cell-mediated TGF-β responses to casein in 26 children with positive OFC to cow’s milk, when compared with 12 age-matched controls with FPIES to other foods. Further study is needed to validate these findings and investigate whether other protein-specific TGF-β responses may be useful in predicting FPIES reaction to additional foods.

Though skin and serum food-IgE testing are negative in a majority of FPIES cases, concomitant IgE sensitization to inciting foods is seen in some FPIES patients and is referred to as “atypical FPIES” [42]. Children with atypical cow milk-FPIES tend to have a more prolonged course, suggesting that cow milk-IgE may play a role in pathogenesis. Recent studies suggest higher rate of atypical FPIES than previously reported; with 24 % of 160 patients in a referral population in the USA [43•] and 33 % of 52 patients in referral population in the UK [26] with IgE sensitization to the inciting food.

Despite publications dating back to the 1940s on FPIES and prevalence similar to IgE-mediated cow’s milk allergy [9], FPIES diagnosis remains delayed in most infants and multiple reactions prior to diagnosis with FPIES are common [13, 14]. Children often undergo extensive evaluations prior to diagnosis which frequently include sepsis work-up, abdominal imaging and surgical consultation; in extreme cases, intestinal biopsy and laparotomy have been performed [13, 14, 25]. Diagnosis can be elusive for a number of reasons including hours of separation in time from exposure to symptom onset; lack of classic respiratory and cutaneous symptoms that are typically associated with IgE-mediated forms of food allergy; inciting foods such as grains, vegetables, and fruits that are traditionally thought of as hypoallergenic; and lack of confirmatory diagnostic tests. Recently published studies from referral populations in the UK and the USA confirm that this continues to be an issue, reporting delays in diagnosis from symptom onset ranging from 6 to 12 months [26, 27].

In contrast to FPIES, laboratory work-up may not be necessary for infants with FPIAP. As long as the infant is well appearing and thriving, it is common practice to empirically eliminate common trigger foods from the maternal or infant diet. If elimination is followed by resolution of clinical bleeding within 72 h, a presumptive diagnosis of proctocolitis is often made without confirmatory challenge. Recent studies call into question the soundness of making a presumptive diagnosis based on resolution with elimination of cow’s milk or other suspected allergen. Jang et al. in 2012 reported that of 10 well infants with rectal bleeding who had histologic findings consistent with FPIAP, only 2 had persistence of rectal bleeding without diet change; the other 8 had resolution of bleeding without food elimination and were diagnosed with transient neonatal eosinophilic colitis [44•]. This study suggests that even with endoscopy findings thought to be consistent with FPIAP, many infants may not, in fact, be reacting to food proteins. In addition to findings of Elizur et al. in Israel that only 3 of 14 infants with presumed FPIAP to milk had recurrence of rectal bleeding with food challenge (discussed earlier) [16•], Arvola et al. in Finland found that just 18 % of the 32 infants with presumed diagnosis of FPIAP in the series were found to have recurrence of symptoms at follow-up challenge [45•]. It should be noted that challenges by Elizur et al. and Arvola et al. were delayed by a mean of 3 and 1 month, respectively; thus, negative challenges may reflect resolution of FPIAP over these months, rather than misdiagnosis. These studies do suggest that follow-up challenge, to confirm persistent sensitivity or to assess for resolution, could be considered 1 to 3 months after food elimination to avoid unnecessarily prolonged elimination diet.

For infants with systemic symptoms, abnormalities on physical exam, upper gastrointestinal symptoms in association with rectal bleeding, or other findings which may be inconsistent with FPIAP, diagnostic work-up is warranted and the clinician should suspect other etiologies for rectal bleeding. Alternative diagnosis to consider in an infant with bloody, mucus stools should include anal fissure, necrotizing enterocolitis, intussusception, infectious colitis (parasitic, bacterial, or viral), Meckel’s diverticulum, coagulopathy, and gastrointestinal duplication cyst.

For infants with FPIAP who have undergone more extensive work-up, laboratory findings are usually mild, and may include anemia, peripheral eosinophilia, and hypoalbuminemia [32, 33, 46]. Colonoscopy and biopsy are generally not performed, but may be indicated when there is diagnostic confusion. These typically show patchy areas of erythema and friability of colonic mucosa, at times with focal erosions and nodularity indicative of focal nodular hyperplasia [32, 33]. Eosinophilic infiltration (along with plasma cells and lymphocytes) in lamina propria and epithelium is evident in the great majority of patients, 89 % from a recent meta-analysis [32]. However, fecal smears are typically negative for eosinophils [6]. Lucarelli et al. in 2011 found APT to be helpful in identifying those infants who required extensively hydrolyzed formula due to persistent symptoms despite elimination of the most common inciting foods. However, this study was small (14 infants) and without controls for comparison; further study is need to validate these findings [47].

In FPE, laboratory work-up and endoscopy with biopsy are necessary to confirm the diagnosis and to differentiate this condition from the many other causes of failure to thrive and diarrhea in infants. The differential diagnosis is similar to chronic FPIES and includes infection, celiac disease, and pancreatic insufficiency (secondary to cystic fibrosis or other conditions), among others. Laboratory studies suggest malabsorption of minerals, proteins, fats, and fat-soluble vitamins. Blood work reveals moderate anemia, hypoproteinemia, and prolonged coagulation time [7, 17•, 18]. Findings from stool studies may include abnormal d-xylose testing (due to carbohydrate malabsorption) and increased fecal fat excretion in approximately 80 % of patients [17•]. Jejunal biopsies show varying degrees of villous atrophy with crypt hyperplasia [7, 48]. In contrast to FPIES and FPIAP, endoscopy may be warranted to confirm diagnosis and lack of villous injury suggests an alternative diagnosis [17•]. Some studies have suggested utility of atopic patch testing; however, these studies have not been validated, and notably, these studies did not utilize endoscopy to confirm the diagnosis [49].

Inciting Foods

In FPIES, cow’s milk, soy, rice, and oat are the most common culprits. Other inciting foods include other grains (barley, corn); meat and poultry (beef, chicken, turkey); egg white; vegetables (white potato, sweet potato, squash, string bean); fruit (tomato); legumes (peanut, green pea, lentil); seafood (fish, crustaceans, mollusks, usually with onset in older children and adults); and the probiotic Saccharomyces boulardii [3, 20, 42, 50–57]. Recently published studies and case reports have added tree nut (type not specified in the study), mushrooms, and egg yolk to this list [24•, 58, 59]. A recent report of an infant reacting to trivial amounts of rice protein on a wrapper suggests that threshold dose for reaction can be small in some patients [60]. Infants with FPIES are often reactive to multiple foods; up to half of infants with solid food FPIES react to more than one food, and approximately one third of infants with cow’s milk-FPIES react to soy and vice versa [43•]. In adult-onset FPIES, the most common trigger foods are seafood followed by egg. Other trigger foods reported included mushroom, corn, peanut, tree nut, and poultry [30].

Breastfeeding is thought to play a protective role in FPIES, and until 2012, FPIES in breastfed infants had not been reported in the literature. Recently published studies note that though rare, chronic FPIES may occur in exclusively breastfed infants due to passage of food proteins such as cow’s milk and soy through breast milk [27, 43•, 61•, 62].

Regarding FPIAP, cow’s milk is similarly the most common inciting food. In contrast to FPIES, infants are commonly exclusively breastfed at time of presentation and improve with removal of inciting food from the maternal diet. From a recent meta-analysis, which included 32 articles and 312 infants, 49 % of infants with proctocolitis were breastfed, 44 % developed proctocolitis on cow’s milk formula, and 6.8 % were on soy [32]. In a study including 95 breastfed infants, stools normalized with sequential elimination of potential triggers from the maternal diet as follows: 65 % improved with elimination of cow’s milk, 19 % with egg, 6 % with corn, and 3 % with soy [5].Five percent of infants were sensitized to multiple foods [5]. There are reports of infants with persistent symptoms despite removal of the above-mentioned foods; these infants’ stools have improved on extensively hydrolyzed formula, suggesting that there are likely other inciting foods in FPIAP that have not yet been identified [47, 63].

Regarding FPE, the most common trigger is cow’s milk or soy, though rice, poultry, fish, and shellfish have also been reported as triggers [17•, 64]. FPE has not been reported in exclusively breastfed infants.

Etiology and Pathophysiology

Immune mechanisms giving rise to FPIES and FPIAP are not well understood. There is strong evidence supporting the involvement of food allergen-specific suppressor CD8 T cells in FPE. See Table 2 for comparison of the etiology and pathophysiology of these three disorders.

Table 2 Etiology and pathophysiology of food protein-induced enterocolitis (FPIES), allergic proctocolitis (FPIAP), and enteropathy (FPE)
Full size table

Management

Acute management of FPIES, FPIAP, and FPE differ depending on severity of illness. Long-term management centers around avoidance of suspected trigger food with interval challenge to assess for resolution.

Acute management of FPIES reactions primarily involves rehydration. Oral rehydration at home may be appropriate for mild reactions if fluids are tolerated by mouth. If there is protracted emesis, severe lethargy, and signs of hypotension or decreased oxygen delivery, resuscitation requires vigorous intravenous hydration in a medical facility. Additional management may include intravenous methylprednisolone (dosed at 1 mg/kg), and supportive therapies such as supplemental oxygen, vasopressors, and methylene blue when necessary. Two recently published case series with five patients each showed rapid improvement in symptoms with ondansetron (intravenous or intramuscular) [65, 66]. Given the efficacy suggested by these studies and the favorable side effect profile, ondansetron may be a useful addition to routine management of FPIES reactions. Epinephrine is not helpful as a first-line drug for managing abdominal pain or vomiting in FPIES and is not routinely prescribed, unless the patient has concomitant IgE sensitization to food.

Long-term management of FPIES involves strict avoidance of inciting foods, provision of an emergency treatment plan in the event of acute reactions, and interval physician supervised food challenge (OFC) to assess for resolution. For infants with cow or soy-FPIES, breast milk or extensively hydrolyzed casein-based formulas (eHCF) are appropriate substitutes, though some may require an amino acid-based formula. In the case of cow’s milk-FPIES, soy has been avoided as a substitute as previous studies reported concomitant soy sensitization in up to 60 % of cases. However, recent publications from a referral population in New York and an Israeli cohort report concomitant soy sensitization in 0 to 37 % of infants with milk-FPIES [9, 43•]. After allergy testing and OFC to soy, soy milk may be an appropriate and economical substitute for many of these infants, especially after 6 months of age. Experts recommend delayed introduction of and possible OFC to common culprits (cow’s milk, soy, grains, legumes, and poultry) if they have not yet been introduced into the diet, although this recommendation is not evidence-based and is best left at the discretion of a treating physician [67–69].

Given that many infants will develop tolerance to the offending food in the first years of life, experts recommend that OFC be performed every 18 to 24 months to assess for resolution. Challenges should be performed under physician supervision, with intravenous access in place, in a monitored setting with capability for resuscitation. Prior to undertaking the challenge, the patient should be assessed for IgE-mediated sensitization with specific IgE and skin testing, as conversion to immediate-type IgE-mediated hypersensitivity to the trigger food has been observed.

In contrast to FPIES, acute management of illness in FPIAP is generally not necessary, as infants with FPIAP are generally well despite mucous or bloody stools. Management centers around elimination of potential inciting foods from the diet. For formula-fed infants, this starts with changing the formula. As some infants will also react to soy, an extensively hydrolyzed formula may be a better option than soy formula. For breastfed infants, continued breastfeeding should be encouraged with empiric elimination of cow milk from the maternal diet. If bloody stools persist, other common triggers for FPIAP may be sequentially eliminated from the maternal diet. Some portion of infants will continue to have bloody stools despite elimination of most common trigger foods and will require hypoallergenic formula in place of breastfeeding: In 7 % of cases, infants will require extensively hydrolyzed formula and 5 % of infants will require amino acid-based formula [33, 70].

Typically, re-challenge with the suspected offending food in FPIAP is not attempted until after 1 year of age [15, 33], although some experts advise 4 to 6 months after elimination. In light of studies discussed above [16,45•], showing tolerance of the suspected inciting food in a majority of infants 1 to 3 months after diagnosis, clinicians may consider shorter duration of food avoidance. For infants on amino acid-based formula, HCF can be introduced for a few months prior to introducing regular milk. A recent study of 30 infants by Baldassarre et al. found that addition of probiotic in addition to allergen avoidance resulted in faster resolution of symptoms compared with allergen avoidance alone [70]. Given that symptoms usually improve fairly quickly and that infants are generally well appearing, this therapy may not be warranted unless further study suggests that this intervention also decreases time to resolution of sensitivity to the inciting food.

Management of FPE is based on studies published years ago, and there is again a paucity of literature in this area. The suspected trigger food should be avoided, and infants should be started on an extensively hydrolyzed formula. As infants are often quite ill on presentation with significant malabsorption and failure to thrive, patients may need intravenous fluids and parenteral nutrition, sometimes for weeks [17•]. Infants may be re-challenged after 1 year of age as a majority of infants develop tolerance by this age [7, 17•].

Prognosis

The prognosis of FPIES, FPIAP, and FPE is generally good, with most affected individuals achieving tolerance in the first years of life. Acute FPIES generally resolves in a few hours with rehydration; chronic FPIES patients usually return to good health in a few days to 2 weeks. Most children outgrow FPIES; the overall rate of resolution ranges widely from 50 to 90 % by the age of 6 years and appears to be both food and population dependent [43•, 71•]. Infants with atypical FPIES (concomitant IgE sensitization) generally have a more protracted course and are at risk for development of IgE-mediated food allergy [42, 43•].

Regarding FPIAP, symptoms generally improve within 72 h of food elimination, though stools may take up to 2 weeks to completely normalize [33]. Most infants tolerate the offending protein by 1 year of age [33]. FPIAP does not appear to be associated with later development of other types of food allergy; in prospective studies, all infants were consuming the offending protein at the age of 3 years without reaction [7].

In comparison to FPIES and FPIAP, active symptom-related FPE take longer to resolve. Villous injury and resulting malabsorption can be significant and can take weeks to normalize following elimination of the offending protein [17•]. Long-term prognosis, however, is good; average age of resolution was 55 weeks in one study and tolerance to the offending protein was achieved in most infants by 2 to 3 years of age [7, 72].

Conclusions

FPIES, FPIAP, and FPE generally have a favorable prognosis; however, in a subset of the affected patients, the manifestations are severe and lead to shock in an acute form of FPIES or to failure to thrive in a chronic form of FPIES and in FPE. There is a strong need to better characterize the pathophysiology of these disorders. Without this knowledge, the identification of biomarkers for diagnosis and development of new treatment strategies will not be possible and patient care will not improve.