afas

atsTitscn1rteOwtiUosiv

taticeiehcdtatdm

June 2013 SELECTED SUMMARIES 1561

an option that limits the number of eliminations is thereforecrucial (Pediatr Allergy Immunol 2012;23:412–419; Immu-nol Allergy Clin North Am 2012;32:83–95).

The authors re-demonstrated the value of empiric milkelimination in a much larger sample than previously. Aswas initially noted in their 2007 study, milk skin andpatch testing have poor NPV (difficulty in differentiatinga true negative from a false negative; J Allergy Clin Im-munol 2007;119:509 –511). Moreover, a positive effect ispredicted for empiric milk-only elimination in addition to atailored diet compared with the effect for other dietarystrategies. Their data also suggest that empiric eliminationof egg, wheat, and soy might be reasonable given theirmodest NPV on allergy testing. Empiric elimination diets(eg, SFED) may be highly successful in both children andadults, although their use is tempered by their unknownmechanism of action in non–food allergic individuals andlack of comparison with other options through any random-ized, controlled study (Clin Gastroenterol Hepatol 2006;4:1097–1102; Gastroenterology 2012;142:1451–1459.e1).

Food allergen testing assesses if there is specific immuno-globulin (Ig)E present against food protein. It directly as-sesses for specific IgE bound to local mast cells in the skin.A positive test reflects sensitization only, meaning the pres-ence of allergen-specific IgE. Allergen patch testing assessesfor T cells (cell-mediated immunity) reactive to foods. Skintesting is highly sensitive and generally detects if any specificIgE is present, but it has poor specificity because the presenceof specific IgE can be clinically irrelevant and not result indisease. Thus, negative tests usually have high NPVs (�95%)nd positive tests have poor PPVs (�50%). The precision ofood patch testing in EoE is even less established, althoughvailable data suggest they have strong NPV and may havetrong PPV as well.

From a methods standpoint, this is an observationalnd uncontrolled study. Causal inference and compara-ive efficacy cannot be assessed. In addition, these data areusceptible to a cluster effect and referral center bias.hus, the findings may not be representative of other sites

n the United States. There are no standardized prepara-ions for patch test materials, so these data may differignificantly across other sites despite attempts at repli-ating this group’s protocol (Ann Allergy Asthma Immu-ol 2005;95:336 –343; Gastroenterology 2012;142:1451–459.e1). Furthermore, the clinical protocol used in theetrospective study is not exactly clear. Although the au-hors do not state it, each of the foods must have beenmpirically withdrawn by at least some of the patients.therwise, the sensitivities of the allergy testing reportedould have all been 100% (no false negatives). In addition,

he use of PPV and NPV are controversial given that theres no precise estimate of the prevalence of EoE in thenited States. In addition, true negative likelihood ratiosr NPVs could not be defined in this study because not allubjects experienced patch testing (if a skin test was pos-tive, no patch test was performed). Therefore, the isolated

alue of patch testing is difficult to assess from these data.

The important take-home message from this study ishat several dietary options may potentially work. Thisllows providers to better select a therapy that best suitsheir patient to maximize the likelihood of compliance. Its our hope that this study will lead to future randomized,ontrolled trials that will help to clarify the specific ben-fits of these different dietary options. Retrospective stud-es of different dietary approaches, like this or Hendersont al (J Allergy Clin Immunol 2012;129:1570 –1578), areelpful for hypothesis generation, but we cannot infer aausal relation from these studies and define a superiorietary approach. A multicenter, randomized, controlledrial is necessary to more properly determine an answer. Inddition, full knowledge of the precision of food allergyesting in EoE will be unknown unless complete testing isone on all patients enrolled, including patch test place-ent even when skin prick tests are positive.

MATTHEW GREENHAWTUniversity of Michigan Food Allergy Center

Division of Allergy and Clinical ImmunologyDepartment of Internal Medicine

University of Michigan Medical SchoolAnn Arbor, Michigan

JOEL H. RUBENSTEINVeterans Affairs Center for Clinical Management Research

Ann Arbor, Michigan andDivision of Gastroenterology

Department of Internal MedicineUniversity of Michigan Medical School

Ann Arbor, Michigan

PROINFLAMMATORY WHEAT ATTACKSON THE INTESTINE: ALPHA-AMYLASETRYPSIN INHIBITORS AS NEW PLAYERS

Junker Y, Zeissig S, Kim SJ, et al. Wheat amylase trypsininhibitors drive intestinal inflammation via activationof Toll-like receptor 4. J Exp Med 2012;209:2395–2408.

Dietary factors play a major role in many gastrointestinaldiseases such as celiac disease, inflammatory bowel diseases(IBD), or gastrointestinal cancers. Mechanisms of actionhave remained mostly unclear. It is believed that variousdietary factors might exert either detrimental (ie, proinflam-matory) or beneficial (ie, anti-inflammatory) effects on thegastrointestinal immune system. Consumption of wheat,barley, or rye causes small intestinal inflammation in celiacdisease patients. It has long been assumed that certain wheatcomponents might also affect innate immunity.

Junker et al of Schuppan’s group now demonstrate thatmembers of the non-gluten alpha-amylase/trypsin inhibitor(ATI) family, which are enriched in wheat and related cereals,are potent activators of various innate immune cells such asdendritic cells (DCs) and macrophages. In their studies,pepsin/trypsin (PT)-digested gliadin caused a dose-depen-dent secretion of interleukin (IL)-8, tumor necrosis factor,and monocyte-chemotactic protein-1 in the human mono-

cytic cell lines THP-1 and U937. Cytokine production was

9

btfwc

AkttI(dwCmoasfaagoa

gcCbdA

swumdb[adis

Ddkfrs1fa1sff(tcmsra

ncpositssot1

fgshfttfcgptwasmg(dsitb

1562 SELECTED SUMMARIES GASTROENTEROLOGY Vol. 144, No. 7

not caused by the �- and �-gliadin fractions, which make up0% of total gliadin, but rather by its �-fraction. Further

analyses of the �-fraction revealed a unique 15-kD proteinand in the Coomassie staining. Mass spectroscopy charac-erized this protein as wheat ATI containing mainly the ATIamily members CM3 and 0.19. These identified ATIs areater-soluble proteins, which constitute the primary mole-

ules of cereals to defend pests and parasites.In further experiments, both untreated and PT-digested

TIs purified from wheat induced proinflammatory cyto-ines in human DCs. Importantly, in all these studies pro-einase K digestion abrogated its cytokine stimulatory po-ential, ruling out lipopolysaccharide contamination.ncubation of human DCs with blocking Toll-like receptorTLR)-4 and CD14 antibodies before addition of ATI re-uced IL-8 secretion, suggesting that ATIs directly interactith TLR4. To further prove specificity of their findings,M3 and 0.19, the main ATI family members detected byass spectrometry, were recombinantly expressed in eukary-

tic HEK-293 cells to guarantee correct protein folding andvoid bacterial contamination. Indeed, both purified ATIstimulated TLR4-MD2-CD14-transfected but not untrans-ected HEK-293 cells, confirming their TLR4-stimulatingctivity. Reduction of disulfides of wheat extracts completelybolished this cytokine stimulatory property of ATIs, sug-esting that the highly disulfide-linked secondary structuref ATIs is necessary to activate TLR4 and that indeed CM3nd 0.19 are the main stimulants of TLR4.

Further in vivo studies showed that only water-solubleliadin caused an up-regulation of duodenal IL-8, monocyte-hemotactic protein-1, and IL-1� after oral ingestion in57BL/6J mice. Importantly, IL-8 transcripts were found toe increased only in wild-type but not TLR4- or MyD88-eficient mice. When mice were gavaged specifically withTIs again, an increase in transcripts of IL-8, IL-1�, and IL-6

was observed. In biopsies from patients with celiac disease,PT-digested gliadin or purified ATIs induced increased IL-8expression. To conclude, the authors identified ATIs as newwheat components, which drive proinflammatory cytokinesynthesis in a TLR4-dependent manner, not only in variousinnate immune cells, but also in duodenal tissues of normalmice and of patients with celiac disease. These data supportthe notion that wheat products might not only mediateceliac disease, but also contribute to other chronic inflam-matory disorders.

Comment. Wheat gluten contains largely water-insolubletorage proteins, such as gliadins and glutenins, as well asater-soluble protein components such as salt-soluble glob-lins, including ATIs. Wheat consumption has been so farainly linked with disorders like wheat allergies or celiac

isease. Although celiac disease reflects a disorder that isased on a genetic predisposition (human leukocyte antigenHLA]-DQ2 and HLA-DQ8 positivity) with a predominantdaptive immune response, there has been substantial evi-ence in the past that innate immune responses are also

nvolved. This is further supported by the fact that only a

mall proportion of people expressing HLA-DQ2 and HLA- e

Q8 develop celiac disease. Innate immune responses eluci-ated by certain wheat components might therefore play aey role in initiating adaptive immune processes. Gliadinractions were shown to induce a proinflammatory cytokineesponse and an increase in the expression of certain adhe-ion molecules in monocytes and DCs (J Immunol 2004;173:925–1933; J Clin Immunol 2007;27:201–209). Such gliadinractions also up-regulate certain chemokine receptors andffect intestinal permeability (Gastroenterology 2008;135:94–204). Despite many attempts, the gliadin fractions re-ponsible for those innate immune activations could not beurther characterized. The findings by Junker et al thereforeor the first time identify wheat components beyond gliadinsie, ATIs), which drive innate and inflammatory responses inhe gastrointestinal tract, although the authors could notonfirm gliadins as innate immune triggers. Importantly,acrophages and DCs, rather than epithelial cells, re-

ponded to wheat ATIs via TLR4, thereby inducing theelease of proinflammatory cytokines/chemokines and initi-ting an inflammatory response.

Establishing that ATIs induced immune activation as aovel mechanism of intestinal inflammation might haveonsiderable translational consequences in our field. First,atients with nonspecific gastrointestinal complaints with-ut definite evidence of celiac disease commonly demon-trate unspecific inflammation in duodenal histologic spec-mens, which so far could not be explained and/or be relatedo any specific pathology. It might well be that, in thoseubjects, their symptoms are driven by ATIs. Second, severaltudies in patients with irritable bowel syndrome (IBS) with-ut evidence of celiac disease have demonstrated that pa-ients improve on a gluten-free diet (Gastroenterology 2001;21:1329–1338; Am J Gastroenterol 2012;107:1908–1912).

Although clinical strategies to avoid gluten-containingoods in these patients are not widely accepted in academicastroenterology, the findings by Junker et al support such atrategy. Patients with non-celiac gluten sensitivity do notave celiac disease, but their symptoms improve on a gluten-

ree diet as demonstrated in a recent study (Am J Gastroen-erol 2011;106:508–514). In this important clinical trial, pa-ients with IBS who were initially asymptomatic on a gluten-ree diet were rechallenged either by placebo or gluten (likelyontaining significant amounts of ATIs). Patients exposed toluten got worse within 1 week, experiencing more bloating,ain, and tiredness. The authors concluded, “Non-celiac glu-en intolerance may exist, but no clues to the mechanismere elucidated.” The findings by Junker et al may providen explanation as to why patients with IBS/non-celiac glutenensitivity may benefit from this dietary restriction and

ight also offer an explanation for the “no man’s land ofluten sensitivity” proposed and discussed by Verdu et alAm J Gastroenterol 2009;104:1587–1594). Gluten may in-uce symptoms similar to functional bowel disorder-likeymptoms, and this may take place via ATIs attacking thentestine. In diarrhea-dominant IBS patients, anti-tissueransglutaminase antibodies and HLA-DQ2 expression haveeen observed in 37% versus 39% of patients (Clin Gastro-

nterol Hepatol 2007;5:844–850). After 6 months of a glu-

tintocl(Ge1lhacrfopc4W2oipAcaidmwhlfoa

a

June 2013 SELECTED SUMMARIES 1563

ten-free diet, stool frequency and gastrointestinal symptomsreturned to normal values in 60% of IBS patients, suggestingthat IBS characterized by this profile most likely responds to agluten-free diet (Clin Gastroenterol Hepatol 2007;5:844–850).

Non-celiac wheat sensitivity (WS) is a recently recog-nized and described disease entity driven by so far unknownwheat components (Am J Gastroenterol 2012;107:1898–1906). In a double-blind, placebo-controlled challenge, theseauthors studied 276 patients with WS and observed 2 clin-ical entities: WS alone versus WS with multiple food aller-gies. Interestingly, WS patients showed a higher rate ofanemia, weight loss, self-reported wheat intolerance, andatopy. The main histologic alteration in the duodenum inWS patients has been an eosinophilic infiltration (Am JGastroenterol 2012;107:1898–1906). It remains to be estab-lished, however, whether ATIs are able to induce this type ofinflammation in patients with WS. If indeed ATIs are in-volved in the pathogenesis of WS, this could also suggestthat ATI-induced intestinal inflammation is able to lead tosystemic symptoms such as anemia or weight loss.

Several authors have suggested an association betweenceliac disease and IBD (J Am Acad Dermatol 1994;31:1050–1051). Whereas the prevalence of IBD in celiac disease seemsto be increasing, the prevalence of celiac disease in IBD iscomparable with controls (Scand J Gastroenterol 2007;42:1214–1220). Anti-Saccharomyces cerevisiae antibodies areknown to be positive in about 65% of Crohn’s disease pa-tients. In most patients, anti-Saccharomyces cerevisiae antibod-ies disappeared during a gluten-free diet, especially in apediatric population (Eur J Gastroenterol Hepatol 2006;18:75–78). Interventional studies with gluten- (and thus ATI-)free diets are currently lacking both for ulcerative colitis andCrohn’s disease.

A balanced diet maintains intestinal mucosal homeostasisin health. Lack of certain phytochemicals and unsaturatedfatty acids can result in an imbalance in immune homeosta-sis, leading to inflammation and pathology (Nat Rev Immu-nol 2012;12:696–708; N Engl J Med 2012;366:181–182). Sev-eral anti-inflammatory dietary signals have been recentlyidentified (Science 2011;334:1561–1565; Cell 2011;147:629–640). These reports demonstrated that certain cruciferousvegetables, and broccoli and cabbage in particular, are phys-iologic ligands of the anti-inflammatory aryl hydrocarbonreceptor and thereby manipulate the host’s immune systemby protecting against inflammation.

The findings by Junker et al have opened a new andfascinating box. The challenge that lies ahead is now toidentify whether these ATIs indeed mediate certain humandiseases such as non-celiac gluten sensitivity. Controlledclinical studies using ATI-free (ie, gluten-free) diets in pa-tients with non-celiac gluten sensitivity are now eagerlyawaited. Several other important questions arise: Are ATIs adominant mechanism in patients with celiac disease? Dothey contribute to other chronic inflammatory disorders inthe gut, such as ulcerative colitis, Crohn’s disease, or otherimmune-mediated diseases? Do they play a role in certainextraintestinal disorders? Would IBD patients benefit from

ATI-free diets? Which diagnostic tools can be developed? o

How can the effects of ATIs be blocked? Only via specificATI-free diets? Or also by specific fermentation or neutral-ization? Evidence is accumulating that the “pathogenicity ofa Western diet” might be influenced by the imbalance of pro-and anti-inflammatory dietary factors. In a sense, ATIs canbe considered as new proinflammatory dietary components.Exciting times are ahead of us in the field of diet andgastrointestinal disorders.

HERBERT TILGROBERT KOCH

ALEXANDER R. MOSCHENDepartment of Internal Medicine I

Department of Gastroenterology, Endocrinology & MetabolismMedical University Innsbruck

Innsbruck, Austria

Reply. We thank Tilg et al for their thoughtful commen-ary on our paper describing wheat alpha-amylase/trypsinnhibitors (ATIs) as activators of intestinal innate immu-ity. We believe that the identification of ATIs as nutri-ional activators of Toll-like receptor (TLR)4 prominentlyn macrophages and dendritic cells may have far reachingonsequences. Although low-level intestinal TLR4 stimu-ation seems to be necessary for normal gut integrityAm J Physiol Gastrointest Liver Physiol 2005;288:G1055–1065), its overstimulation contributes to intestinal and

xtraintestinal pathology (Biochem Soc Trans 2007;35:473–1478). Usually, the gut is protected from bacterial

ipopolysaccharide, the classical TLR4 ligand, which isydrolyzed and inactivated by gastric acid and intestinallkaline phosphatase (J Exp Med 2012;209:2395–2408). Inontrast, ATIs are largely resistant to proteolysis andeach the intestinal innate immune system in vivo. There-ore, although being less potent than lipopolysacchariden a molar basis, ATIs can provide a continuous moderateroinflammatory stimulus in our wheat-based diets. Theoncentration of ATIs in modern wheat is high, reaching%–5% of total wheat protein (Proteome Sci 2011;9:10).ith an average daily consumption of approximately

00 g of wheat in the United States (http://en.wikipedia.rg/wiki/Wheat) and a wheat protein content of approx-

mately 10%, this amounts to nearly 1 g of ingested ATIser day, an amount that is 3-fold higher in the near East.s enzyme inhibitors, ATIs comprise a family of �16ompact proteins serving as major defense moleculesgainst fungi and parasites. As such, their content hasncreased steadily via resistance breeding, even in the lastecades. Thus, the old diploid variants such as Triticumonococcum contain no or little ATI protein and activity,hereas these have increased steeply in modern hexaploidigh yielding variants (Proteome Sci 2011;9:10; unpub-

ished data). Notably, despite the presence of similar de-ense proteins in other staples such as corn or rice, it isnly the ATIs of gluten containing cereals that stronglyctivate TLR4 (unpublished data).

It can be hypothesized, and preliminary supportive datare emerging, that nutritional ATIs may contribute not

nly to classical intestinal inflammatory diseases, but also