|Home | About | Journals | Submit | Contact Us | Français|
Eosinophils are multifunctional leukocytes that increase in various tissues in a variety of disorders. Locally, they can be involved in the initiation and propagation of diverse inflammatory responses. In this review, the clinical association of eosinophils with diseases of the skin, lung and gastrointestinal tract is summarized. An approach to determining the causal role of eosinophils in these diseases is presented. Recent findings concerning molecular diagnosis, etiology and treatment are discussed.
Eosinophils are multifunctional leukocytes implicated in the pathogenesis of numerous inflammatory processes including infections (parasitic helminths, bacterial and viral), non-specific tissue injury, malignancy, and allergic diseases.1 In response to a variety of stimuli, eosinophils are recruited from the circulation into the tissue where they modulate immune responses through muliple mechanisms. Triggering of eosinophils by cytokines, immunoglobulins, and complement can lead to the release of an array of proinflammatory cytokines, such as chemokines, interleukins (e.g. IL-2, IL-4, IL-5, IL-10, IL-12, IL-13, IL-16, IL-18), transforming growth factor (TGF)-α/β, lipid mediators such as platelet-activating factor (PAF) and leukotriene (LT)C4, free radicals, and mitochondrial DNA. These molecules have proinflammatory effects that include upregulation of adhesion systems, modulation of cellular trafficking, regulation of vascular permeability, mucus secretion, and smooth muscle constriction. In addition, eosinophils can initiate adaptive immunity by acting as antigen-presenting cells and secreting T helper cell chemokines. Furthermore, eosinophils can serve as major effector cells inducing tissue damage and dysfunction by releasing cytotoxic granule proteins, inflammatory lipid mediators and mitochondrial DNA.1 In this chapter, we summarize the association of eosinophils with diseases involving three tissues, the skin, respiratory tract, and gastrointestinal (GI) tract. Focusing on clinical data, we discuss differential disease diagnosis, therapy and pathogenesis. A more detailed discussion of disease mechanisms and the detailed results of early eosinophil-targeted novel therapy are provided in another review in this issue.2
Eosinophil infiltration is found in a broad spectrum of skin disorders (Table 1).3 It is a characteristic feature of allergic diseases or parasitic infestations, but it is also observed in autoimmune diseases, hematologic diseases, as well as in association with tumors, and bacterial or viral infections. Depending upon the disease, eosinophils can be the predominant cell infiltrate such as in eosinophilic cellulitis or can be part of a mixed inflammatory infiltrate in the dermis such as in eczematous reactions. Eosinophils may infiltrate the epidermis presenting as eosinophilic spongiosis in particular in autoimmune bullous diseases, insect bite reactions or acute contact dermatitis. Eosinophil infiltration of the deep dermis and subcutaneous fat tissue can be observed in eosinophilic cellulitis, parasitic infections, erythema nodosum, vasculitis, or lymphomas. Peripheral blood eosinophilia may be associated with tissue eosinophilia, e.g. in drug reactions with eosinophilia and systemic symptoms (DRESS), atopic dermatitis, or bullous pemphigoid.
In hematoxylin and eosin stained skin specimens, eosinophils are noticeable as round shaped cells stuffed with coarse eosinophil granules. In subacute and chronic eczematous lesions, disrupted oval shaped eosinophils may also be found. Extracellular deposits of granular proteins can be detected in varying amounts either as separate little granules or as a thin coating on collagen bundles. The latter are called flame figures and can typically be seen in eosinophilic cellulitis. Immunofluorescence staining using antibodies directed against eosinophilic cationic protein (ECP) or major basic protein (MBP) allows a more sensitive detection of eosinophils and extracellular granular protein depositions compared with hematoxylin and eosin staining.
Eosinophils do not enter the skin under physiological states. Mechanistically, cutaneous eosinophilia can be from a primary problem internal to the eosinophil or may be caused by stimuli outside the cell.3 In either case, increased production, recruitment and/or survival of eosinophils is likely. Hematologic disorders affecting multipotent or pluripotent hematopoietic stem cells may involve the eosinophil lineage. In these diseases, mutations that represent intrinsic defects in eosinophils cause eosinophil proliferation and tissue infiltration including the skin. Cutaneous manifestations are described as multiple erythematous papules, plaques and nodules, or generalized erythematous maculopapular eruptions often associated with pruritus. By means of cytogenetic and molecular techniques, a number of diseases formerly defined as idiopathic hypereosinophilic syndrome (HES) can now be classified as separate entities. Clonal eosinophilia is often associated with rearrangements involving the genes of the platelet-derived growth factors A and B (PDGFRA, PDGFRB) resulting in increased tyrosine kinase activity.4 Notably, patients with HES due to the fusion of the PFGFRA and FIP1L1 genes respond to imatinib therapy.4
More commonly, extrinsic eosinophilic disorders are observed, in which skin eosinophilia is due to cytokine release by either T cells or tumor cells. Cytokines involved in the development of skin eosinophilia include interleukin (IL-)-3, IL-5 and granulocyte/macrophage colony-stimulating factor (GM-CSF). The expression of IL-5 in association with eosinophilic skin disorders has been reported in atopic dermatitis,5, 6 exanthematous drug reactions, 7 urticaria,8 episodic angioedema with eosinophilia,9 bullous pemphigoid,10 eosinophilic fasciitis,11 eosinophilic folliculitis,12 cutaneous T cell lymphoma,13 eosinophilic cellulitis14 and HES with skin involvement.15 IL-3 expression has been detected in blister fluids of bullous pemphigoid.16 In Langerhans cell histiocytosis17 as well as in atopic dermatitis, atopy patch test reactions and cutaneous late phase reactions, the expression of both IL-3 and GM-CSF has been shown.16, 18 Expression of the chemokine eotaxin has been observed in atopic dermatitis,19 drug reactions,20 autoimmune-blistering diseases like dermatitis herpetiformis and bullous pemphigoid,21 parasitic dermatoses,22 eosinophilic folliculitis,12 but also lymphomas, (e.g. cutaneous T cell lymphoma) and Hodgkin’s disease.23 The functional role of eosinophils in the pathogenesis of skin diseases remains largely unknown. Depending on the skin disease, a role in host defense, immunoregulation and/or remodeling and fibrosis can be assumed. Specific cutaneous eosinophilic disorders are described below.
As the name implies, Wells syndrome is characterized by an intense infiltration of eosinophils, extracellular granule deposition and flame figures in the dermis. Patients present with recurrent episodes of acute pruritic dermatitis, seldom with blisters, painful edematous swellings or persistent urticarial eruptions.24 An increased expression of IL-5 has been reported in a number of cases. The cause is not known, but some patients developed eosinophilic cellulitis in association with hematological disorders, infections, or anti-TNF-alpha therapy. Corticosteroids are usually helpful in Wells syndrome. Eosinophilic cellulitis may also occur as a cutaneous manifestation of HES. Other skin manifestations of HES are erythematous macules, papules or nodules, blisters, necrosis, ulcerations, purpura, lichenoid eruptions or urticarial lesions, and pruritus. The skin is affected in 37% of HES patients.25 Anti-IL-5 antibody therapy was shown to improve skin symptoms in HES patients.26 In a subgroup of patients with HES, IL-5 producing clonal T cells has been identified. These T cells often exhibit an abnormal phenotype in as far as they have an either higher, lower or absent expression of lineage-associated markers. Those patients usually present with cutaneous symptoms.27
EPF presents as annular clusters of sterile follicular papules and pustules predominantly on the face and trunk that heal with postinflammatory hyperpigmentation but tend to recur periodically.28 The histology shows a dense follicular and perifollicular infiltrate of eosinophils and scattered lymphocytes, and sometimes follicle destruction. The classic type of EPF affects immunocompetent subjects. Meanwhile two other subtypes of EPF have been identified. Infancy-associated EPF often involves the scalp. More commonly, EPF is seen in context with immunosuppression. EPF has been reported in association with infections, in particular AIDS, medications, autoimmune diseases, as well as autologous peripheral blood stem cell and allogeneic bone marrow transplantation. A pathogenic role for eosinophils in response to fungi (Malassezia), demodex mites and bacteria has been suggested.
Despite various cutaneous and histopathological presentations of drug reactions (e.g. maculopapular rashes, erythema multiforme, acute generalized exanthematous pustulosis, pseudolymphomatous and granulomatous drug reactions), the presence of eosinophils in the skin is a quite striking finding.29 DRESS, also named hypersensitivity syndrome, presents with an acute, severe skin eruption that may develop from a maculopapular rash into erythroderma, as well as with fever, lymphadenopathy, hepatitis, blood eosinophilia, and other organ involvement due to hypereosinophilia.30 Eosinophils accompanied by other inflammatory cells are found in the skin and the lymph nodes. Severe hepatitis, in which eosinophilic infiltration or granulomas as well as hepatocyte necrosis and cholestasis are striking features, may result in liver failure accounting for the high mortality rate of 10%. The treatment is based on high-dose corticosteroids. Drugs known to cause DRESS are anticonvulsants, sulfa drugs, antimicrobial agents, anticancer drugs, nonsteroidal anti-inflammatory drugs and anti-diabetic agents.
Tissue eosinophilia is a typical feature of AD. The numbers of eosinophils in the skin are usually modest (2.8 cells/mm2 [range 0 to 90.3]) and correlate with disease severity, as well as the degree of spongiosis in acute exacerbations and marked epidermal hyperplasia in chronic stages.31 Besides eosinophils, eosinophil-derived products such as ECP, EDN (also abbreviated as EDX), and MBP are present in increased amounts in the blood and the skin of AD patients. The measurement of ECP in serum is frequently used as a tool for monitoring AD activity and response to therapy.32 Immunostaining with antibodies to MBP and ECP has demonstrated that eosinophil granule proteins are not only present inside eosinophils but also in the extracellular spaces, suggesting eosinophil degranulation. In biopsies obtained from chronic AD lesions, intact eosinophils are located predominantly within the perivascular mononuclear cell infiltrate. In contrast, in the upper dermis extensive extracellular MBP deposition are observed in the near absence of intact eosinophils.33 The presence of mostly disrupted eosinophils has been confirmed by electron-microscopy studies, which revealed various degrees of eosinophil degeneration ranging from intact eosinophils with granule abnormalities, to intact eosinophils with abnormal granules and pseudopod-like extensions, to eosinophils with degenerated cell and/or nuclear membranes to free eosinophil granules.34 Improvement of AD upon both systemic and topical therapy is usually associated with a decrease of eosinophils and other inflammatory cells in the skin. However, the administration of an anti-IL-5 antibody showed only moderate effects on clinical symptoms although blood eosinophils were almost completely depleted.35
Bullous pemphigoid (BP) is due to an autoimmune response to structural components of junctional adhesion complexes leading to damage of the dermal-epidermal junction with subepidermal blister formation.36 Autoreactive B and T cell responses against the hemidesmosomal antigens BP180 and BP230 have been identified. IL-5 as well as eotaxin are abundant in blister fluids and the production of IL-5 is associated with blood eosinophilia and significant eosinophil infiltration in the skin of BP patients.37 Eosinophils are thought to be implicated in blister formation by releasing toxic granule proteins (ECP, MBP) and proteolytic enzymes, however, the molecular mechanisms are not well understood. In dermatitis herpetiformis, a specific cutaneous manifestation of gluten-sensitive enteropathy due to anti-tissue transglutaminase antibodies, a neutrophilic infiltrate undermingled with eosinophils is found in the papillary dermis. The expression of eotaxin in lesional skin and increased levels of serum ECP suggest a role for eosinophils in disease pathology.21
In Langerhans cell histiocytosis, among the infiltrate of Langerhans cells, scattered or clusters of eosinophils can be found in the papillary and deeper dermis, respectively.38 Langerhans cells produce a broad spectrum of proinflammatory cytokines, e.g. GM-CSF, and chemokines, and are thus likely to recruit and activate eosinophils directly or via stimulation of other cell types. Predominant T helper 2 type cytokine production by cutaneous T cell lymphoma results in eosinophilia, extracellular granule protein depositions as well as increased IL-5 levels in the skin and/or peripheral blood.39 Currently, it is not known whether the eosinophils modulate the proliferation of the malignant cells.
Eosinophils are relatively rare in the normal lung and so they stand out both in tissue and airway lumen samples when present in increased numbers. A number of lung diseases are associated with blood and tissue eosinophilia (Table 2). The extent to which eosinophils cause tissue damage in these diseases remains controversial, but most evidence points to them as being pro-inflammatory effector cells in non-infectious disorders where they are prominent. The most common association, at least in industrialised countries, is with asthma and related airways diseases. Research into these conditions has resulted in much of our current understanding of the role of eosinophils in lung disease. These findings, as well as a summary of other eosinophil-associated lung conditions, are now discussed in some detail.
There has been active debate about the role of eosinophils in asthma for at least the last four decades. Initially they were regarded as ameliorative cells able to dampen inflammatory responses, but this changed with the work of Gleich and colleagues who demonstrated that eosinophil basic proteins, particularly MBP, were toxic for bronchial epithelial cells and MBP was present in large amounts in the airways of patients who had died from asthma.40 At the same time controlled studies of bronchoscopy in mild asthma demonstrated that a BAL eosinophilia tracked with active disease.41 An assumption was made that the abnormal physiology that defines asthma (airway hyperresponsiveness [AHR]) and variable airflow obstruction, was secondary to the airway eosinophilia. This hypothesis was underpinned by a persuasive paradigm that emerged in the 1990’s that asthma was driven by activation in the bronchial mucosa of Th2 lymphocytes which through the generation of IL-4, IL-5 and IL-13 are closely associated with a blood and tissue eosinophilia both in humans and animal models.42–44 However, subsequent studies of induced sputum that allowed a more thorough examination of the relationship between airway inflammation and asthmatic airway dysfunction revealed at best a very weak correlation between the degree of airway eosinophilia and the abnormal physiology. Conditions such as eosinophilic bronchitis (EB), which presents as chronic cough, were identified in which airway eosinophilia occurred in the absence of airway dysfunction and some asthmatics were identified who had airway dysfunction without eosinophilia.45–47 Eosinophils are therefore neither necessary nor sufficient for an asthma like airway dysfunction to be present.
Monoclonal antibodies against IL-5 have provided insight into the potential role of eosinophils in asthma. The drug which has been investigated in greatest detail is mepolizumab.48 The first clinical efficacy study of mepolizumab was in a human allergen challenge model. This demonstrated a marked reduction in blood and sputum eosinophils without any effect on either the early or late response or AHR. Although this was a small study the interpretation of which was open to debate,49 a larger trial of clinical asthma came to the same conclusion that eosinophils were not responsible for either the symptoms or physiological abnormalities which characterise asthma.50 In this double blind placebo controlled study, 362 patients with moderate asthma were given three injections of drug one month apart. There was no difference in symptoms or lung function between the two groups. There was about a 50% reduction in severe exacerbations (SE) in the 750 mg group, but the study was not powered to look at exacerbations and this did not quite reach significance p<0.061). Of note, evidence of eosinophilic disease was not an entry criteria for the study and sputum analysis was only undertaken in 37 patients. Another caveat to the conclusion that eosinophils are not causal in asthma pathophysiology is that while mepolizumab was very good at reducing blood and sputum eosinophils it only had a modest (50%) reduction in bronchial wall eosinophils.50 Interestingly, this is similar to the effect of systemic steroids in refractory asthma which have a marked effect on luminal eosinophils, but a much less marked effect on tissue eosinophils (51 and personal observation Andrew Wardlaw). The reasons for this are not clear, but suggest that IL-5 is more important in tissue based eosinophil migration whereas migration through the vascular endothelium into tissue is relatively IL-5 independent with chemotactic stimuli perhaps being more important. This would be consistent with the observation that in contrast to two dimensional Boyden chamber like assays, in a 3-dimensional model of tissue based trafficking, migration in response to growth factors was more robust than migration in response to eotaxin.52
Do these clinical trials mean that eosinophils are not acting as effector cells in asthma? Two issues need to be considered. First, in the above studies inclusion was not restricted to patients with eosinophilic disease. In patients with hypereosinophilic disease, many of whom had respiratory pathology, mepolizumab was highly effective in allowing a reduction in the dose of glucocorticoids required to control disease activity.53 Second, it depends what is meant by asthma. Clarifying this question requires a nuanced understanding of the pathophysiology of asthma and related airways diseases. Asthma was originally defined as marked variability in airflow obstruction over short (minutes) periods of time either spontaneously or in response to treatment and this correlates very closely with the presence of AHR. Asthma symptoms of wheeze, chest tightness and shortness of breath are closely related to this phenotype. However, over the last few decades definitions of asthma have become more descriptive and attempt to encompass the complete physiology seen in asthma and other airway diseases.54 With the new insights gained by very targeted treatments such as mepolizumab there is a strong case to be made for returning to the original definition.55 It is helpful in this regard to deconstruct airway disease into its component pathophysiological abnormalities, a process that is aided by an A to E classification system.56 In this scheme A stands for the Asthma/AHR phenotype, B for Bronchitis, C for Cough, D for Damage (fixed airflow obstruction, bronchiectasis, emphysema) and E for Extrapulmonary factors such as adherence, co-morbidity and triggers. Eosinophils are found in the bronchitis phenotype and are most closely associated with the clinical phenotype of SE although they are also associated with cough particularly in conditions such as EB. As noted above, the majority of people dying of asthma (the end result of untreated SE), have marked eosinophilic inflammation.57, 58 Eosinophils were the best predictor of SE in a steroid reduction study and approaches to management targeting the airway eosinophilia resulted in a marked reduction in SE.59, 60 It therefore follows that the patients who are most likely to respond to anti-eosinophilic therapy are those with marked eosinophilic airway inflammation who suffer from SE. This was confirmed in a study of twelve months treatment with mepolizumab in patients with refractory eosinophilic asthma with SE as the primary outcome measure where a significant 40% reduction in SE was observed.61 Predictably there was no effect on lung function, AHR or day to day asthma symptoms although there was an improvement in quality of life reflecting the impact of SE on morbidity. Interestingly, there was no effect on exhaled nitric oxide, dissociating this biomarker from eosinophilic inflammation and clinical improvement. These findings were echoed in a study published at the same time of patients with marked eosinophilic airway disease (several did not have AHR), using a steroid reduction design in which mepolizumab was found to be highly effective at reducing exacerbations. A marked and significant reduction in steroid dose was possible in the active compared to the placebo group.62 These studies represent direct evidence that eosinophils are causal in the pathogenesis of asthma associated SE. An important outcome of these findings is that eosinophilic airway disease (which will respond to anti-eosinophilic therapy), is a distinct albeit overlapping phenotype from asthma as defined by Hargreave and Nair.62 Studies of anti-eosinophil agents therefore need to have inclusion criteria and outcome measures relevant to the eosinophilic Bronchitis phenotype rather than the Asthma/AHR phenotype (variable airflow obstruction, AHR and symptoms). These patients are clinically recognisable in an airways disease clinic and can also be identified as an eosinophil inflammation predominant cluster in an objective multi-dimensional analysis of asthma phenotype.63 These studies also give some insight into the pathogenesis of asthma SE.64 As noted above, the major effect of mepolizumab was to reduce the luminal eosinophilia with a modest effect on tissue eosinophils. It suggests therefore that SE are principally a luminal event with obstruction caused by blockage of bronchi with mucus and cell debris rather than smooth muscle mediated bronchoconstriction. This would be consistent with the pathology of asthma mortality where mucus impaction is often the primary pathological cause of death. It would also explain why SE are often relatively bronchodilator resistant. Focusing on luminal events in relation to SE would also be consistent with the well-established importance of viral infections in triggering SE, presumably through an interaction with the epithelium in the context of eosinophilic inflammation.65
One of the more common causes of lung eosinophilia is fungal colonization. The most florid expression of this is allergic bronchopulmonary aspergillosis (ABPA), the primary diagnostic criteria of which are asthma, proximal bronchiectasis, a positive skin prick test and positive specific IgE and IgG to Aspergillus fumigatus (Af) and a high total IgE.66 A blood eosinophilia and positive sputum culture for Af are secondary criteria. Classically patients present with exacerbations of their asthma characterised by fleeting lung shadows, but this presentation is now unusual possibly because of the increased use of potent high dose inhaled steroids. A marked significant blood and tissue eosinophilia help guide diagnosis. A number of other molds can cause an ABPA-like picture most of which are other Aspergillus or Penicillium species and IgE sensitisation to a range of fungi is common in more severe asthma.67, 68 Two studies have shown that treatment with anti-fungals is beneficial in ABPA although both were small and the benefits modest.69, 70 Relatively few patients with asthma and a positive specific IgE to Af fulfill all the criteria for ABPA although whether Af plays a significant part in their asthma in these circumstances is not clear. Interestingly, patients with severe asthma and fungal sensitisation had a significant improvement in quality of life after treatment with itraconazole although this class of drugs increases endogenous steroid levels which is a confounding factor.71
Eosinophilic pneumonia is an uncommon condition characterised by a marked peripheral blood and tissue/BAL eosinophilia and pneumonic lung shadowing. Patients may present acutely with breathlessness, hypoxia, general malaise and sometimes fever and can be very unwell. They respond dramatically to high dose systemic steroids.72 Occasionally, presumably because of rapid trafficking of eosinophils into the lung, the peripheral blood can be in the normal range. In cases such as drug allergy or parasitic infection where there is an identifiable cause recurrence can usually be avoided, but many patients with idiopathic disease often develop chronic eosinophilic pneumonia with recurrent relapse requiring long term oral steroids. Relapse of eosinophilic pneumonia is often of slow onset and associated with airflow obstruction in the absence of wheeze or bronchodilator responsiveness suggestive of small airways disease. Relapses invariably respond well to high dose oral steroids.
Churg-Strauss syndrome (CSS) is a rare condition in which patients have asthma (which is often severe), a marked peripheral blood eosinophilia and evidence of a multi-system vasculitic disorder.72, 73 Upper airway symptoms are common and mononeuritis multiplex, which is often slow to respond to treatment, is a characteristic feature. The cause is unknown and the extent to which eosinophils are directly responsible for the tissue damage is unknown (trials of mepolizumab in this condition have not yet been reported). It is interesting however, that neural damage is such a feature considering the neurotoxic properties of EDN and the neurotrophic properties of eosinophils.74, 75 Treatment of the acute phase consists of potent immunosuppressants including high dose systemic steroids, cyclophosphamide and azathioprine although in the longer term low dose oral steroids are often sufficient to maintain remission. There is an association between treatment with leukotriene receptor antagonists and the onset of CSS but this is generally considered to be coincidental although a causal relationship has not been ruled out. 76
Tobacco smoking associated COPD is often a disease that is contrasted with asthma with a more neutrophilic and CD8+ T cell-associated pathology. However, just as some asthma is non-eosinophilic quite a lot of COPD is characterised by eosinophilic airway inflammation particularly during exacerbations.77 These patients are often more steroid responsive than their non-eosinophilic counterparts.78
A number of helminthic parasites have a larval stage that passes through the lung and causes respiratory symptoms.79, 80 The exact pattern of illness varies with the parasite with various acronyms (Table 2). Generally the condition looks like a combination of asthma and eosinophilic pneumonia with low grade fever, dry cough, chest discomfort, shortness of breath, wheeze and occasionally haemoptysis. The chest X-ray classically shows lung shadowing that can be pneumonic in appearance and often lasting from a few days to weeks. Parasites can sometimes be identified in lung tissue samples. Tropical pulmonary eosinophilia caused by infection with filarial parasites can produce a more chronic illness with lethargy and anorexia in association with asthma like symptoms and occasionally mediastinal lymphadenopathy, cavitation and pleural effusion .
Eosinophils are present throughout the healthy GI tract, except for the esophagus, which typically contains no eosinophils.1 Eosinophil-associated gastrointestinal disorders (EGID) are characterized by a high level of eosinophils within isolated or multiple segments of the GI tract. Over the past decade, there has been a striking increase in the incidence of primary EGID, as well as a robust increase in data linking the development of EGID to atopy. The most common form of EGID is eosinophilic esophagitis (EE), which is characterized by a relatively high level of eosinophils within the esophagus (without an acid-induced etiology).81 Other forms of EGID include: eosinophilic gastritis, eosinophilic enteritis, eosinophilic colitis, and eosinophilic gastroenteritis. These disorders are less frequent than EE, but their pathogenesis and treatment are somewhat similar. Clinically, patients with EGID often have failure to thrive (especially in pediatrics), dysphagia, vomiting, abdominal pain and/or diarrhea.82 Patients with EE often present with symptoms that mimic gastroesophageal reflux disease (GERD). The number of eosinophils per high power field (HPF) required to diagnose EGID has not been uniformly agreed upon, which can make the diagnosis of EGID challenging. For this reason, an experienced pathologist knowledgeable with the number of eosinophils typically found in the GI tract at their medical center is essential for interpreting biopsy specimens in patients with suspected EGID. Agreed upon criteria for the diagnosis of EGID is currently being pursued. The diagnostic criteria for EE, established at the First International Gastrointestinal Eosinophilic Research Symposium, is >15 eosinophils/HPF noted in at least one of four biopsy specimens when GERD is ruled out.81 The quantity and distribution of GI eosinophils in pediatric patients without apparent pathologic disease has been reported,83 though the application of this data to the diagnosis of EGID has not been established. In addition to the elevated eosinophil numbers, the usual morphology of the GI tract is typically disrupted in EGID. Depending on the location of the eosinophil accumulation, additional manifestations may include the presence of eosinophilic microabscesses, disruption of the surface epithelium cryptitis, basilar hypertrophy or lamina propria fibrosis.84 Although the presence of elevated eosinophils and disruption of the typical GI tract morphology are crucial for the diagnosis of EGID, the isolated presence of GI eosinophilia is not sufficient. The differential diagnosis for GI eosinophilia is broad and includes EGID, HES, collagen vascular disease, inflammatory bowel disease, and parasitic or fungal infection; disease differentiation based on primary or secondary causes is useful (Table 3). In order to diagnose EGID, the patient must have a biopsy and clinical presentation consistent with EGID; other causes of GI eosinophilia (e.g. parasitic infection, drug hypersensitivity) must be ruled out. This review will focus on primary EGID, disorders not associated with known causes for the eosinophilic inflammation.
The esophagus normally does not contain eosinophils so the finding of esophageal eosinophils denotes pathology. In addition to EE, many disorders are accompanied by eosinophil infiltration in the esophagus, such as GERD, eosinophilic gastroenteritis, GERD, parasitic and fungal infections, inflammatory bowel disease, HES, esophageal leiomyomatosis, myeloproliferative disorders, carcinomatosis, periarteritis, allergic vasculitis, scleroderma and drug injury.85 EE is classified into primary and secondary with the primary subtype including the atopic, nonatopic and familial variants and the secondary subtype including one composed of systemic eosinophilic disorders (e.g. HES) and the other of non-eosinophilic disorders (Table 3). Primary EE has also been called idiopathic EE or allergic esophagitis. The sibling recurrence risk ratio has been estimated to be over 50-fold86 and the familial form of EE is noted in about 10% of patients.87
The cause of EE is poorly understood, but food allergy has been implicated. In fact, most patients have evidence of food and aeroallergen sensitization as defined by skin prick and/or allergen-specific IgE tests; however, only a minority have a history of food anaphylaxis.88 Evidence suggests that esophageal eosinophilic inflammation is mechanistically linked with pulmonary inflammation based on the finding that delivery of specific allergens or the Th2 cytokine IL-13 to the lungs of mice induces experimental EE.89, 90 Increased eosinophil levels in the esophagus of patients with seasonal allergic rhinitis with hypersensitivity to grass has been published.91 Recent studies have found a strong relationship between atopy and EE.92–94 Indeed, patients with EE commonly report variable seasonal symptoms. In addition to eosinophils, T cells and mast cells are elevated in esophageal mucosal biopsies, suggesting chronic Th2 associated inflammation.95 Exposure to antigen via an epicutaneous route primes for marked eosinophilic inflammation in the esophagus triggered by only a single respiratory antigen exposure.96 IL-5 is required for this eosinophilic inflammation, suggesting a Th2-dependent mechanism. Notably, overexpression of IL-5 induces EE and neutralization of IL-5 completely blocks allergen or IL-13-induced EE in mice.89, 90, 97
A landmark advance in EE research is the recent genome wide microarray profile analysis of esophageal tissue.98 Investigators compared gene transcript expression in esophageal tissue of patients with EE, chronic esophagitis (typical of GERD), and normal individuals. Notably, dysregulated expression of ~1% of the human genome led to the identification of an EE genetic signature. Interestingly, eotaxin-3 was the most overexpressed gene in EE patients and levels correlated with disease severity; a finding that has now been independently replicated.99 Furthermore, a single nucleotide polymorphism (SNP) in the eotaxin-3 was overrepresented in EE patients compared with control individuals. Interestingly, mice with a genetic ablation of the eotaxin receptor (CCR3) were protected from the development of experimental EE. Notably, eotaxin-3 has been shown to be produced by esophageal epithelial cells, and induced by the Th2 cytokine IL-13, which is also markedly overexpressed in the esophagus of EE patients.100 Taken together, these results strongly implicate eotaxin-3 in the pathogenesis of EE and offer a molecular connection between Th2 inflammation and the development of EE.
A specific food allergen avoidance trial is often indicated for patients with atopic EE, and if unsatisfactory or practically difficult (when patients are sensitized to many allergens), a diet consisting of an elemental (amino acid based) formula is often advocated. Interestingly, it has been shown that an elemental diet has potential to reduce the number of eosinophils in the esophageal biopsies and improve symptoms in patients with primary EE (allergic or non-allergic sub-types).101 Patients on elemental diets frequently require a surgically placed gastrostomy tube in order to achieve adequate caloric support. Glucocorticoids (systemic or topical) have also been used with satisfactory results. Systemic glucocorticoids are used for acute exacerbations, while topical glucocorticoids are used to provide chronic control. A study that followed EE patients for 10 years supports the efficacy of continuing corticosteroids and food elimination therapy for EE.101 When using topical steroids we recommend the use of a metered-dose inhaler without a spacer and having the patient swallow the medicine to promote deposition on the esophageal mucosa. The toxicity associated with inhaled glucocorticoids (e.g. adrenal suppression) is unlikely to be seen with swallowed fluticasone due to first-pass hepatic metabolism following gastrointestinal absorption. However, about 10% of patients treated with topical fluticasone develop esophageal candidiasis. While the metered-dose inhaler is recommended with the topical fluticasone, another study has shown success of using an oral suspension of budesonide in EE patients who were unable to use inhalers.102 In the first placebo controlled double blind trial in EE, swallowed topical fluticasone was demonstrated to be effective in inducing disease remission including reduction in eosinophil, mast cell, and CD8 T cell levels, as well as the degree of epithelial hyperplasia.103 Notably, only half of the patients responded to topical glucocorticoids, and 10% responded to placebo. A recent study also showed the promising effect of anti-human-IL-13 antibody in an animal model of IL-13-induced airway and esophageal eosinophilia. 104 In addition, in two open-label trials, humanized anti-IL-5 monoclonal antibody therapy has been shown to be helpful in small numbers of patients.105, 106 Larger scale trials are currently underway. Finally, even if GERD is not present, neutralization of gastric acidity with proton pump inhibitors may improve symptoms and the degree of esophageal pathology.
In contrast to the esophagus, the stomach and intestine have readily detectable baseline eosinophils under healthy conditions, confounding the diagnosis of EGID involving these tissues. For the purposes of this review, eosinophilic gastritis, enteritis and gastroenteritis are grouped together because they are similar clinically and because there is a lack of information available concerning their pathogenesis; however, it is likely that they are indeed distinct processes in most patients. These diseases are characterized by the selective infiltration of eosinophils in the stomach and/or small intestine with variable involvement of the esophagus and/or large intestine. It is now appreciated that many diseases are accompanied by eosinophilia in the stomach, such as infection (parasitic and bacterial) including Helicobacter pylori, periarteritis, inflammatory bowel disease, allergic vasculitis, HES, myeloproliferative disorders, scleroderma, drug injury and drug hypersensivity.82 Similar to EE, these disorders are classified into primary and secondary subtypes. The primary group includes the atopic, nonatopic and familial variants while the secondary subtype contains two groups, one composed of systemic eosinophilic disorders (e.g. HES) and the other of non-eosinophilic disorders (Table 3). Primary eosinophilic enteritis, gastritis, and gastroenteritis have also been called idiopathic or allergic gastroenteropathy. Primary eosinophilic gastroenteritis involves multiple disease entities subcategorized into different types based on the level of histological involvement: mucosal, muscularis and serosal forms.107 Of note, either layer of the GI tract can be involved; as such, endoscopy and biopsy can be normal in patients with the muscularis and/or serosal subtypes.
While these diseases are idiopathic, an allergic mechanism has been suggested.108 Indeed, elevated total IgE and food-specific IgE has been detected in most patients. On the other hand, focal erosive gastritis, enteritis and occasionally esophagitis with prominent eosinophilia, such as the dietary (food) protein-induced enterocolitis and dietary protein enteropathy are characterized by negative skin tests and absent specific IgE.109 Most patients have positive skin prick tests to a variety of food antigens, but do not have typical anaphylactic reactions, consistent with a delayed-food hypersensitivity.
In clinical studies, increased production of Th2 associated cytokines (e.g. IL-4 and IL-5) by peripheral blood T cells has been reported in patients with eosinophilic gastroenteritis.108 Furthermore, lamina propria T cells derived from the duodenum of patients with EGID preferentially secrete Th2 cytokines (especially IL-13) when stimulated with milk proteins.110 IgA deficiency has also been associated with eosinophilic gastroenteritis; perhaps this could be related to the associated increased rate of atopy or to an occult gastrointestinal infection in these patients. Eosinophilic gastroenteritis and the dietary protein-induced syndromes (enterocolitis, enteropathy and colitis) may represent a continuum of EGID with similar underlying immunopathogenic mechanisms. In addition, eosinophilic gastroenteritis can frequently be associated with protein-losing enteropathy.111
Eliminating the dietary intake of the foods implicated by skin tests (or measurement of allergen-specific IgE levels) has variable effects, but complete resolution is generally achieved with elemental diets.111 Once disease remission has been obtained by dietary modification, the specific food groups are slowly reintroduced (at ~3 week intervals for each food group) and endoscopy is performed every three months, to identify disease status. Drugs such as sodium cromoglycate, montelukast, mycophenolate mofetil (an inosine monophosphate dehydrogenase inhibitor), ketotifen, suplatast tosilate and “alternative Chinese medicines” have been suggested, but are generally not successful. However, a publication reported a successful long-term remission of eosinophilic gastroenteritis following montelukast treatment.112 In our institution, an appropriate therapeutic approach includes a trial of food elimination if sensitization to food is found by skin tests and/or measurement of specific IgE levels. If no sensitization is found or if specific food avoidance is not feasible, elemental formula feedings are initiated.
The management of eosinophilic gastroenteritis, in addition to an amino acid based diet, includes the following: systemic and topical steroids, non-corticosteroid therapy, and management of other EGID complications (such as iron deficiency and anemia).113 Anti-inflammatory drugs (systemic or topical steroids) are the main therapy if diet restriction has failed or is not feasible. There are several forms of topical glucocorticoids designed to deliver drugs to specific segments of the GI tract (e.g. budesonide tablets [Entocort™ EC] designed to deliver drug to the ileum and proximal colon). In severe cases refractory or dependent upon glucocorticoid therapy, intravenous alimentation or immunosuppressive antimetabolite therapy with azathioprine or 6-mercaptopurine are alternatives. Finally, even if GERD is not present, neutralization of gastric acidity with proton pump inhibitors may improve symptoms and the degree of esophageal and gastric pathology.
Eosinophils accumulate in the colon of patients with a variety of disorders, including infection (pin and dog hookworms), eosinophilic gastroenteritis, allergic colitis of infancy, drug reactions, and vasculitis, (e.g. CSS, and inflammatory bowel disease).114 Dietary protein-induced proctocolitis of infancy syndrome (also known as allergic colitis of infancy), is the most common cause of blood in the stools in the first year of life.115 Similar to other EGID, these disorders are classified into primary and secondary (Table 3) with the primary type including the atopic and nonatopic variants composed of systemic eosinophilic disorders (e.g. HES) and the other by non-eosinophilic disorders.
Eosinophilic colitis is usually a non-IgE-associated disease. In fact, some studies point to a T-lymphocyte-mediated process, but the exact immunologic mechanisms responsible have not been identified.116 It has been reported that allergic colitis of infancy might be an early expression of protein-induced enteropathy or protein-induced enterocolitis syndrome. Cow’s milk and soy proteins are the most frequently implicated foods in allergic colitis of infancy, but a variety of food proteins can also provoke the disease. Interestingly, this condition may more commonly occur in infants exclusively breast-fed and can even occur in infants fed with protein hydrolysate formulas.
Treatment of eosinophilic colitis varies primarily depending upon the disease subtype. Eosinophilic colitis of infancy is generally a benign disease. Upon removal of the offending protein in the diet, the gross blood in the stools typically resolves within days, but occult blood loss may persist longer. Treatment of eosinophilic colitis in older individuals usually requires medical management since IgE-associated triggers are usually not identified. Drugs such as montelukast, sodium cromoglycate, and histamine receptor antagonists are typically not successful. Anti-inflammatory drugs including aminosalicylates and systemic or topical glucocorticoids appear to be efficacious, but careful clinical trials have not been conducted. There are several forms of topical glucocorticoids designed to deliver drugs to the distal colon and rectum, but eosinophilic colitis usually involves the proximal colon. In severe cases, alternative therapy includes intravenous alimentation or immunosuppressive therapy with azathioprine or 6-mercaptopurine.
Eosinophilic tissue diseases are a heterogeneous group of diseases which include common conditions such as asthma and atopic dermatitis, less common but regularly diagnosed diseases such as EE, as well as rare diseases such as EP and CSS. The eosinophilia in these diseases may be associated with allergy to common aeroallergens, but include rarer causes of eosinophilia such as drug allergy and (in non-industrialised countries) parasitic infection. However, in many patients with eosinophilic tissue disease the cause remains elusive. Eosinophilic tissue disease is almost invariably highly responsive to glucocorticoids which are the mainstay of treatment. True steroid resistance is rare although in refractory diseases where the inflammation is peripheral, systemic as opposed to topical steroids are often required. Apparent steroid resistance in eosinophilic disease should always raise questions about adherence to treatment. Although usually effective, systemic steroid therapy is limited by toxicity, providing the impetus for better anti-eosinophil drugs. Promisingly anti-IL5 strategies appear to be well tolerated and effective and provide strong clues about the role of eosinophils in various tissue diseases. To optimally use these drugs we need to recognise that eosinophilic tissue diseases have distinct features. For example, in the case of the lung, asthmatic patients that may benefit from anti-eosinophil directed therapy have a selective phenotype including marked blood and/or sputum eosinophilia, nasal polyposis, and patterns of recurrent severe steroid responsive exacerbations. Taken together, emerging concepts have been presented which highlight the seriousness of eosinophil-associated tissue diseases, the potential role of eosinophils in these processes, and appropriate approaches to differential diagnosis and therapy.
This work was supported in part by the NIH NIAID, FAAN, Food Allergy Project, CURED Foundation, and Buckeye Foundation (to MER).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.