In Vitro Gliadin Challenge: Diagnostic Accuracy and Utility for the Difﬁ Cult Diagnosis of Celiac Disease

nature publishing group ORIGINAL CONTRIBUTIONS 111

In Vitro Gliadin Challenge: Diagnostic Accuracy and Utility for the Difﬁ cult Diagnosis of Celiac Disease

Raffaella Tortora , MD 1 , Ilaria Russo , PhD1 , Giovanni D. De Palma , MD2 , Alessandro Luciani , PhD3 , Antonio Rispo , MD1 , Fabiana Zingone , MD1 , Paola Iovino , MD4 , Pietro Capone , MD1 and Carolina Ciacci , MD4

OBJECTIVES: Diagnosis of celiac disease is difﬁ cult when treatment with gluten-free diet (GFD) is started before diagnosis and / or when the results of tests are inconsistent. The objective of this study was to evaluate the in vitro gliadin challenge.

METHODS: The study cohort included patients without celiac disease (negative controls, n = 57), patients with

celiac disease (positive controls, n = 166 untreated and n = 55 on GFD), and patients with diffi cult COLON/SMALL BOWEL diagnosis (n = 59). All patients underwent endoscopy for collection of duodenal samples, which served for the diagnosis of celiac disease and for the in vitro evaluation of the gliadin-induced mucosal expression of seven infl ammatory markers: PY99, ICAM-1 (intercellular cell adhesion molecule), HLA-DR, CD3, CD25, CD69, and transglutaminase 2 IgA. Diagnostic work-up for celiac disease included the search of specifi c serum antibodies. Patients of the diffi cult diagnosis group were asked to stop GFD for repeated search of these antibodies under untreated conditions. The area under the receptor-operated curve (ROC) was used for statistical analyses on accuracy.

RESULTS: HLA-DR had the highest accuracy for celiac disease diagnosis in analyses on negative controls and positive controls also excluding patients on GFD (area under ROC = 0.99). Accuracy of test did not increase combining data of HLA-DR with data of other markers. Findings were similar in the 39 patients of the difﬁ cult diagnosis group undergoing the search celiac disease-speciﬁ c antibodies under untreated conditions.

CONCLUSIONS: The in vitro response of mucosal HLA-DR to gliadin is an accurate tool for the diagnosis of celiac disease also in patients with difﬁ cult diagnosis. Am J Gastroenterol 2012; 107:111–117; doi: 10.1038/ajg.2011.311; published online 27 September 2011

INTRODUCTION gluten into the patient’ s diet for an appropriate period. Also, diag- Celiac disease is a digestive disorder that damages the small intes- nosis of celiac disease can be diffi cult when the search of serum tine and interferes with the absorption of nutrients from food. antibodies and the duodenal biopsy results are not concordant. Patients with celiac disease cannot tolerate gluten-rich cere- Genetic markers as the HLA DQ2– DQ8 genes are of help only in als and, in particular gliadin, which is the toxic fraction of the excluding celiac disease as their presence is common also in non- gluten of wheat (1 ). Diagnosis of celiac disease is based on the celiac subjects ( 5,6 ). evidence of high levels of serum auto-antibodies, including anti- In the presence of gliadin, the duodenal mucosa of celiac patients tissue transglutaminase antibodies (a-tTG) or anti-endomysium undergoes several modifi cations that, in part, can be reproduced antibodies (EMA) and on the evidence of mucosal damage in the in vitro (7,8 ). Th e short-term in vitro exposure of the duodenal duodenal biopsy ( 2 –5 ). Diagnosis of celiac disease can be diffi cult mucosa of celiac patients to gliadin induces crypt hyperplasia, vil- in patients who begin treatment with a gluten-free diet (GFD) lus atrophy, and T lymphocyte recruitment in the lamina propria, aft er non-standardized food intolerance tests and in the absence increasing the number of intraepithelial T lymphocytes ( 9– 11 ). of adequate diagnostic work-up ( 4 ). Once this treatment is Th e in vitro response of the duodenal mucosa to gliadin is gener- initiated, diagnosis of celiac disease requires the reintroduction of ally defi ned as the gliadin challenge, a test which could represent

1 Department of Clinical and Experimental Medicine, Federico II University of Naples, Naples , Italy ; 2 Department of Surgery, Endoscopy Unit, Federico II University of Naples, Naples , Italy ; 3 Institute of Pediatrics, University of Foggia, Foggia , Italy ; 4 University of Salerno, School of Medicine, Gastroenterology, Campus di Baronissi , Salerno , Italy . Correspondence: Carolina Ciacci, MD, University of Salerno, School of Medicine, Gastroenterology, Campus di Baronissi, via S. Allende 84081 Baronissi (SA), Salerno 80131, Italy. E-mail: [email protected] or [email protected] Received 8 November 2010; accepted 14 August 2011

a tool for objective assessment of celiac disease when diagnosis is Total patients diffi cult. Th e same test was used also for the evaluation of immu- 347 nological modifi cations in celiac disease ( 12 ). Th e present study aimed to evaluate the diagnostic accuracy of Incomplete data the in vitro gliadin challenge in patients without celiac disease, N = 10 with celiac disease, and in patients with diffi cult diagnosis.

Study cohort METHODS N = 337 Study design Th e study was divided into two parts ( Figure 1 ). In the fi rst part, Negative controls the accuracy of the gliadin challenge was evaluated in patients in n = 57 whom celiac disease had been excluded (negative controls) and in patients with serology- and biopsy-proved celiac disease (positive Positive controls controls). In the second part, the accuracy of the gliadin challenge n = 221 was evaluated in cases where diagnosis of celiac disease was dif-

COLON/SMALL BOWEL fi cult. Th e study fulfi lled all the items of the STARD Statement for diagnostic accuracy (http://www.stard-statement.org ). Untreated celiac disease Treated celiac disease n = 166 n = 55 Patients Th e study enrolled patients undergoing upper endoscopy at the Federico II Gastrointestinal Unit, which is a tertiary center for food intolerance and celiac disease. Th e inclusion criteria were Difficult diagnosis adult age (≥ 18 years) and informed written consent. Participants n = 59 underwent, a dietary interview for evaluation of treatment with GFD, upper endoscopy for duodenal biopsies, and assessment Figure 1 . Study design. of HLA status ( 13,14), serum a-tTG levels (ELISA, Diamedix, Miami, FL), and EMA (EMA IIF, Eurospital, Trieste, Italy). Th e study included three groups of patients: negative controls, posi- and EMA every 3 months. Celiac disease diagnosis was made dur- tive controls, and diffi cult diagnoses. Negative controls were ing follow-up if celiac disease-specifi c serology was consistently patients without celiac disease as assessed by negative serology positive for both a-tTG and EMA. If the serology was constantly and histology selected to match sex and age of celiac patients of negative during the follow-up, celiac disease was excluded. the positive control group. Th e negative control group included patients with peptic disease (n = 21), gastroesophageal refl ux dis- Gliadin challenge ease (n = 22), irritable bowel syndrome (n = 7), and infl ammatory Duodenal biopsies and tissue culture . Th e in vitro gliadin chal- enteropathies (n = 2 with autoimmune enteropathy, three with lenge was performed by expert biologists (I.R, A.L, L.M) who Crohn ’ s disease, one with acute gastroenteritis, one with Giardia L . were blind to the characteristics of the patients. Eight duodenal infestation). Th e positive controls included celiac disease patients fragments were collected from each patient. Two fragments were under treated and untreated conditions. Treated celiac patients oriented on Whatmann paper for routine histology. Th e remain- were on GFD since at least 1 year aft er complete diagnostic work- ing six fragments were used for gliadin challenge and placed in up including serology and histology. Untreated celiac patients ice-cold tissue-culture medium within 20 min and cultured as were cases with positive serology who underwent upper endos- follows. Fragments were placed on a stainless-steel mesh posi- copy for completion of diagnostic work-up. Th e diffi cult diagnosis tioned over the central well of an organ-culture dish containing group was made of patients with suspected celiac disease in whom culture medium (37 ° C) with the epithelium facing upward. Th e diagnosis could not be made because the diagnostic work-up was gliadin challenge was performed as previously described adding initiated while on treatment with GFD and / or of non-concord- a gliadin digest (1 mg / ml) to four samples ( 9,10 ). Two fragments ant diagnostic tests. Th e list of combination of non-concordant were cultured for 3 h for evaluation of early markers of infl amma- tests included the presence of EMA antibodies with the absence of tion namely: PY99 (anti-phospho-tyrosine-mAb), HLA-DR (the a-tTG or vice versa and the presence of antibodies with the absence major histocompatibility complex class II molecules), ( 14) and of histological abnormalities in the duodenal mucosa. ICAM-1 (intercellular cell adhesion molecule) (7,11 ). Two fragments were cultured for up to 24 h for evaluation of the delayed Follow-up studies markers of infl ammation, namely CD3 (a marker of mature T Aft er completion of the gliadin challenge, participants of the diffi - lymphocyte), CD25 (the inducible interleukin-2 receptor in both cult diagnosis group were prescribed a gluten-containing diet and lymphoid and myeloid cells), and CD69 (a marker of T-cell activa- were followed up for up 1 year with assessments of serum a-tTG tion) (7,11 ). Th e remaining two fragments served as controls and

The American Journal of GASTROENTEROLOGY VOLUME 107 | JANUARY 2012 www.amjgastro.com Difficult Diagnosis of Celiac Disease 113

were cultured for 3 or 24 h in the same way without the addition on villous enterocytes at time point 3 h and on crypt enterocytes of gliadin (blank samples). Incubations were stopped by washing, at time point 24 h, respectively. Intensity score was based on the embedding tissues in Optimal Cutting Temperature (OCT) (Tis- absolute count of positive cells per mm2 of mucosa for ICAM-1, sue TEK, Milews Laboratories, ElKhart, IN), and snap freezing in CD69, CD3, and CD25 (0: < 5 cells, 1: 5 –29 cells, 2: 30– 59 cells, cooled isopenthane. Samples were stored in liquid nitrogen, and 3: ≥60 cells). Intensity score of and TG2 and IgA was based on prepared for both histology and immunofl uorescence analysis. the co-expression of both markers below the basement membrane Aft er the fi nding of the deposits in the mucosa of celiac patients (along the villous or along the crypt or around mucosal vessels) as and potentially celiac patients, ( 15) all samples (stored and in- reported (15 ). For each one of the markers, the challenge was con- coming) underwent also the search for the transglutaminase 2 sidered positive when the intensity score in gliadin-exposed sam- IgA (TG2 –IgA) deposits before and aft er 3 and 24 h ( 15,16 ). ples was at least one unit higher than the intensity score in blank samples. For example, the challenge was considered positive if the Immunoﬂ uorescence intensity score was 0 in the blank sample and 1 or higher in the For analyses of antigen expression and tissue distribution by gliadin-exposed sample, as well as if the intensity score was 1 in indirect immunofl uorescence, 4 μ m cryosections were sepa- the blank sample and 2 or higher in the gliadin-exposed sample, rately incubated in the presence of the following antibodies: etc. Vice versa, the challenge was considered negative if the inten- PY99 (1:80, mouse IgG2b; Santa Cruz Biotechnology, Santa Cruz, sity score was 0 in the blank samples and in the gliadin-exposed

CA); HLA-DR (1:10, Beckton Dickinson, San Jose, CA); ICAM-1 sample, as well as if the intensity score was 1 in the blank sample COLON/SMALL BOWEL (1:400; Ylem Pasteur, Rome, Italy); CD3 (1:100; Dako, Glostrup, and 1 or lower in the gliadin-exposed sample, etc. Denmark); CD25 (1:30; Rabbit Polyclonal; BD Pharmingen, San Diego, CA); and CD69 (1:80; BD Pharmingen) (17 ). Anti- Statistical analysis gen expression and distribution was visualized using a donkey Analysis of variance was used for comparisons of continuous anti-rabbit IgGs conjugated to Alexa Fluor 488 for 60 min at variables among the groups. For each marker, the rate of positive room temperature. Two-color immunofl uorescence with anti- cases in negative controls and of negative cases in positive con- CD25 and anti-CD3 antibodies were performed as previously trols was used to calculate the area under the receptor-operated described (11 ). Isotype control antibodies (IgG1 or IgG2), iso- curve (ROC). type-matched non-immune IgGs, or isotype-matched antibodies against inappropriate blood group antigens were used as control of specicity. Data were analyzed under fl uorescence examination RESULTS using a LSM510 Zeiss confocal laser scanning unit (Carl Zeiss, Descriptive statistics Jena, Germany). HLA-DR expression was graded from absent to Of the 347 participants enrolled in the study, 10 were excluded very strong (0 – 3) ( 11 ). ICAM-1, CD3 / CD25, and CD69-positive because of incomplete data ( n = 4 w i t h s u p e r fi cial biopsy samples, cells were counted per mm2 of mucosa. Th e number of epithelial n = 3 with necrotic samples, and n = 3 with contaminated cul- cells stained with PY99 were counted per 100 epithelial cells. Th e tures). Th us, the study cohort consisted of 337 patients. Table 1 TG2 deposits search was done by double labeling for human IgA reports the descriptive statistics of the three groups. Gender and (green) and for TG2 (red) with the use of monoclonal mouse anti- age had a similar distribution between negative controls and posi- bodies against TG2 (CUB7402; NeoMarkers, Fremont, CA). Th e tive controls. Serum a-tTG averaged in the normal range in nega- sections were washed in phosphate-buff ered saline and incubated tive controls and in treated patients of the positive control group, with a mixture of fl uorescein isothiocyanate-labeled rabbit anti- above the normal range in untreated celiac patients, and in the body against human IgA (1:100; Dako) to detect (in green) IgA borderline range in the diffi cult diagnosis group. deposits and R-phycoerythrin-labeled rabbit anti-mouse antibody (1:40; Dako) to detect (in red) TG2, for 30 min in the dark. Challenge in negative controls and positive controls Finally, the sections were washed several times in phosphate- Figure 2 shows representative examples of the immunofl uores- buff ered saline and mounted by glycerol/ phosphate-buff ered cence experiments in the presence of gliadin and blanks samples saline (1:10) ( 15,18,19 ). in negative controls and positive controls. Table 2 reports for each marker the prevalence of positive cases Assessment of immunofl uorescence intensity . I n t e n s i t y o f (i.e., false-positive cases) in negative controls and of negative cases immunofl uorescence was scored from 0 to 3 for each one of the positive controls (i.e., false-negative cases). PY99 had the second seven markers as reported (8,11,15 ). Intensity score of PY99 was highest rate of false-positive cases. HLA-DR had no false-positive based on the percent of stained cells per 100 epithelial cells for case at both times of incubation and only one false-negative case. (0: < 25 % ; 1: 25 – 49 % ; 2: 50 – 74 % ; 3: ≥ 75 % ). Intensity score of HLA- ICAM-1 and CD69 had the highest rate of false-positive cases and DR was based on the cellular localization of expression (0: no ex- no false-negative case. CD3 and CD25 had high rates of false-posi- pression; 1: expression in basolateral membrane; 2: expression in tive cases and of false-negative cases. TG2– IgA deposits had the basolateral membrane and cytoplasmic compartment; 3: expres- highest rate of false-negative cases. Th e highest value of the area sion in basolateral membrane, cytoplasmic compartment, and under the ROC was found for HLA-DR at both incubation times. apical membrane). Th e evaluation of HLA-DR expression focused Th e area under the ROC of HLA-DR did not signifi cantly increase

Table 1 . Gender, age, anthropometry, serum antibodies, treatment with gluten-free diet (GFD), and HLA DQ2/ DQ8 presence of the 330 participants in the study (mean± s.d. or prevalence)

Negative control group Positive control group Difﬁ cult diagnosis group Untreated celiac disease patients Treated celiac disease patients

Number 57 166 55 59 Women ( % ) 66.7 % 74.1 % 74.5 % 72.9 % Age (years) 34.2 ± 10.2 31.8 ± 10.6 30.4 ± 10.9 36.1 ± 13 Weight (kg) 56.4 ± 7.6 59.7 ± 10.7 55.4 ± 10.5 56.6 ± 8.9 Height (cm) 164.1 ± 9.3 165.6 ± 8.8 162.5 ± 8.8 163.2 ± 6.8 BMI (kg/ m 2 ) 21.3 ± 2.2 21.6 ± 3.1 20.8 ± 2.6 21.1 ± 2.9 a-tTG (U / ml) 3.4 ± 2.5 19.8 ± 5.6 4.9 ± 1.4 6.3 ± 5.4 On GFD 0 0 55 33 Presence of HLA 26 166 55 46 DQ2 / DQ8 COLON/SMALL BOWEL a-tTG, anti-tissue transglutaminase antibodies; BMI, body mass index.

when data of this marker were combined with data of anyone of for repetition of analyses under untreated conditions. Patients with the other markers (not shown). negative response of HLA-DR had a HLA profi le, which was against Analyses were repeated aft er exclusion of treated celiac patients the possibility of celiac disease in 9 out of 11 cases together with to investigate the possible confounding of treatment with GFD. inconsistent results of serology and histology of pre-challenge anal- Area under the ROC for HLA-DR was the highest one also in this yses. Patients with positive response of HLA-DR had a HLA pro- set of analyses (area under ROC = 0.99). Findings for other mark- fi le compatible with celiac disease in all cases and a pre-challenge ers were similar to data of Table 2 with exception of TG2 – IgA. Th e serology compatible with celiac disease in four out of nine cases. area under the ROC of TG2 – IgA increased up to 0.90 aft er exclusion of treated celiac patients. Analyses were done also focusing on blank samples to investi- DISCUSSION gate the accuracy of the expression of these markers without the Th e present study shows that in the gliadin challenge test, HLA- incubation with gliadin. For all markers, the area under the ROC DR has high accuracy for celiac disease diagnosis because of was much lower than values found with the use of the gliadin high specifi city and high sensitivity. Findings for this marker challenge. Th e highest value was found for the area under the were consistent at both incubation times, in the absence and ROC of HLA-DR 24 h (0.71), the lowest value for the area under in the presence of treatment with GFD, and in patients with the ROC of TG2– IgA (0.64). Th e inaccuracy of immunofl uo- diffi cult diagnosis because of non-concordant tests and/ or the rescence assessment in blank samples mainly refl ected a high confounding of GFD initiated before an appropriate diagno- number of false-negative cases. In fact, for all markers, the rate of sis of celiac disease. Findings for the other markers of in vitro positive controls (i.e., celiac patients) with intensity score = 0 was mucosal response to gliadin were less accurate and did not above 50% whereas the rate of negative controls with intensity increase the accuracy of information derived by the evaluation score ≥ 1 was below 16 %. of HLA-DR alone. Th e data from this study indicate the diagnostic accuracy of Gliadin challenge results in difﬁ cult diagnosis group the in vitro gliadin challenge in the identifi cation of celiac dis- Th irty-nine patients of the diffi cult diagnosis group accepted to ease in the event of a diffi cult diagnosis. It has previously been stop GFD aft er the gliadin challenge to undergo a re-assessment shown that the in vitro gliadin challenge, combined with EMA of the celiac disease-specifi c serology. Table 3 reports data about supernatant levels, may be used to maximize the sensitivity of accuracy of markers of the gliadin challenge in these 39 patients both histology and serology (20 – 22 ). In the present study, we divided by presence/ absence of celiac disease-specifi c antibodies observed a clear modifi cation in the expression of markers of in the analyses repeated under untreated conditions. Th e compar- the T-cell immune response following in vitro gliadin challenge ison between these two subgroups tended to give results, which almost exclusively in celiac disease (19,23 ). Indeed, in case of were similar to the results of the comparison between negative celiac disease, the culture of mucosal explants exposed to glia- controls and positive controls. HLA-DR had the highest area din showed specifi c features of infl amed celiac mucosa, such under the ROC also in this set of analyses. as PY99 and HLA-DR expression by epithelial cells ( 13 ), an Table 4 reports data about the HLA-DR response in the gliadin increase in CD3 and CD25 cellular expression, and ICAM-1 challenge and other characteristics in the 20 patients of the diffi - molecules in the lamina propria. We experienced a 3% failure cult diagnosis group who refused to stop the treatment with GFD rate in the ability to cultivate biopsies for the in vitro challenge

The American Journal of GASTROENTEROLOGY VOLUME 107 | JANUARY 2012 www.amjgastro.com Difficult Diagnosis of Celiac Disease 115

Negative controls Positive controls MARKER Blank PT-gliadin Blank PT-gliadin

HLA-DR

PY99

ICAM COLON/SMALL BOWEL

CD69

IgA

TG2/IgA TG2

Merge

Figure 2 . Examples of immunostaining in duodenal samples of negative controls (without celiac disease) and positive controls (with celiac disease) after incubation with gliadin-free medium (blank) and with gliadin-rich medium (PT-gliadin). ICAM, intercellular cell adhesion molecule; TG2/ IgA, transglutaminase 2/ IgA.

with gliadin; however, standardized procedures may lower In our hands, TG2 – IgA deposits had high accuracy only when the percentage of these technical problems (such as obtaining analyses were limited to patients who were not on GFD. It is pos- superfi cial tissue samples). sible that our storage and/ or the culture of samples of duodenal Several patients in the diffi cult diagnosis group were on GFD mucosa have aff ected the TG2– IgA expression and reduced its without any evidence of celiac disease. Th is is now a common accuracy. occurrence and is in accordance with the recent data (24,25 ), Th e in vitro gliadin challenge could be of help in psychological demonstrating that a number of patients who receive a diagnosis support of patients who are without celiac disease but are afraid to of celiac disease on the basis of negative specifi c serum markers return to a gluten-containing diet. and minimal alteration of mucosa at histology are unlikely to have According to previous studies, full appraisal of the immunologi- celiac disease. cal response to gliadin should be conducted and analyzed for a

Table 2 . Percent prevalence of positive cases in negative controls Table 4. Characteristics of the 20 patients with difﬁ cult diagnosis (patients without celiac disease), of negative cases in positive who refused to stop gluten-free diet after the gliadin challenge controls (patients with celiac disease), and derived area under divided in two subgroups by challenge results of HLA-DR the receptor-operated curve (ROC) Pre-challenge False positive False negative HLA-DR Number of Pre-challenge histology Marsh cases in nega- cases in positive Area under ROC challenge patients DQ2 /DQ8 serology classiﬁ cation tive controls (% ) controls (% ) (95 % CI) Negative 6 Absent EMA − 3b Number of 57 221 a-tTG + patients 3 Absent EMA − 2 PY99 16.0 % 1.4 % 0.91 (0.85 / 0.97) a-tTG – HLA-DR 3 h 0.0 % 0.5 % 0.99 (0.99 / 1.00) 2 Present EMA − 2 a-tTG + HLA-DR 24 h 0.0 % 0.5 % 0.99 (0.99 / 1.00) Positive 4 Present EMA + 3a ICAM-1 18.0 % 0.0 % 0.91 (0.85 / 0.97) a-tTG + CD69 18.0 % 0.0 % 0.91 (0.85 / 0.97) 2 Present EMA − 2 a-tTG – CD3 10.5 % 4.5 % 0.87 (0.80 / 0.95) COLON/SMALL BOWEL 3 Present EMA − 3b CD25 3.5 % 6.3 % 0.87 (0.80 / 0.94) a-tTG – TG2 – IgA 0.5 % 6.8 % 0.94 (0.89 / 0.99) deposits a-tTG, anti-tissue transglutaminase antibodies; EMA, endomysium antibodies.

CI, conﬁ dence interval; ICAM, intercellular cell adhesion molecule; TG2– IgA, transglutaminase 2 IgA.

in vitro gliadin challenge might be less an expensive procedure than repeated endoscopies and biopsies. The proposed organ Table 3 . Percent prevalence of positive cases, negative cases and culture limited to these markers can be performed in many area under the receptor-operated curve (ROC) of each marker pathology laboratory equipped for cell culture and immuno- in patients with difﬁ cult diagnosis divided in the subgroup histochemistry. Although in this simplified form the gliadin with positive serology and the subgroup with negative serology challenge should be performed only in tertiary centers for the as assessed in repeated analyses under untreated conditions diagnosis and therapy of celiac disease and strictly limited to (gluten-containing diet) cases in which a GFD or non-concordant tests interfere with Positive cases Negative cases a correct diagnosis of celiac disease. Further investigation is in patients with in patients with needed on the in vitro gliadin challenge in non-celiac gluten- negative sero- positive serology (without logy (with celiac Area under ROC sensitive individuals. celiac disease) disease) (95 % CI) In conclusion, the in vitro gliadin challenge is not only a model for studying the mechanisms of the mucosal immune response Number of 19 20 patients in celiac disease but also a useful tool for reaching a fi nal diagnosis in “ diffi cult ” cases of celiac disease, in which sensitivity to PY99 31.6 % 10.0 % 0.79 (0.64/ 0.94) gluten is suspected but undemonstrated by standard diagnostic HLA-DR at 3 h 15.8 % 0.0 % 0.92 (0.82/ 1.00) tests. HLA-DR at 24 h 15.8 % 0.0 % 0.92 (0.82/ 1.00) ICAM-1 47.4 % 0.0 % 0.76 (0.61/ 0.92) CONFLICT OF INTEREST CD69 47.4 % 0.0 % 0.76 (0.61/ 0.91) Guarantor of the article: Carolina Ciacci, MD. Specifi c author contributions: Conception, planning, study design, CD3 52.6 % 5.0 % 0.71 (0.54/ 0.88) collection and interpretation of data, and draft ing of the article: CD25 47.4% 10.0 % 0.71 (0.54/ 0.88) Raff aella Tortora and Carolina Ciacci; laboratory testing, interpreta- TG2 – IgA 20.0 % 36.8 % 0.72 (0.55/ 0.88) tion of data, and draft ing of the article: Ilaria Russo; diagnostic test- deposits ing: Giovanni D. De Palma; laboratory testing: Alessandro Luciani; CI, conﬁ dence interval; ICAM, intercellular cell adhesion molecule; TG2– IgA, diagnostic testing, interpretation of data, and draft ing of the article: transglutaminase 2 IgA. Antonio Rispo; statistical analysis: Fabiana Zingone; diagnostic testing: Paola Iovino; data collection: Pietro Capone. Financial support: Th is study was supported by the Italian Research number of markers for up to 24 h ( 8 ). On the basis of our find- Program (PRIN 2006) and by Regione Campania, Assessorato ings, we suggest a gliadin challenge with testing of HLA-DR Sanit à . alone and only for 3 h of culture with gliadin. This simplified Potential competing interests: N o n e .

The American Journal of GASTROENTEROLOGY VOLUME 107 | JANUARY 2012 www.amjgastro.com Difficult Diagnosis of Celiac Disease 117

Study Highlights 8 . M a i u r i L , P i c a r e l l i A , B o i r i v a n t M et al. D e fi nition of the initial immuno- logic modifi cations upon in vitro gliadin challenge in the small intestine of WHAT IS CURRENT KNOWLEDGE celiac patients. Gastroenterology 1996 ; 110 : 1368 – 78 . 9 . T r o n c o n e R , G r e c o L , M a y e r M et al. In siblings of celiac children, rectal 3 In the presence of gliadin, the intestinal mucosa of celiac gliadin challenge reveals gliadin sensitization not restricted to celiac HLA . patients undergoes several modifi cations that, in part, can Gastroenterology 1996 ; 111 : 318 – 24 . be reproduced in vitro . 1 0 . P i c a r e l l i A , D i To l a M , S a b b a t e l l a L et al. Multicentre Organ Culture 3 System Study Group. Usefulness of the organ culture system in the in vitro Previous studies have shown that short-term in vitro expo- diagnosis of coeliac disease: a multicentre study . Scand J Gastroenterol sure of the intestinal mucosa of celiac patients to gliadin 2006 ; 41 : 186 – 90 . induces crypt hyperplasia, villus atrophy, and T lymphocyte 11 . Maiuri L , Ciacci C , Ricciardelli I et al. Association between innate response recruitment in the lamina propria, increasing the number of to gliadin and activation of pathogenic T cells in coeliac disease . Lancet intraepithelial T lymphocytes (IELs). 2003 ; 362 : 30 – 7 . 3 The response of the intestinal mucosa to gliadin is gener- 1 2 . M a z z a r e l l a G , S t e f a n i l e R , C a m a r c a A et al. Gliadin activates HLA class I- restricted CD8+ T cells in celiac disease intestinal mucosa and induces the ally defi ned as the gliadin challenge and the results of this enterocyte apoptosis . Gastroenterology 2008 ; 134 : 1017 – 27 . test may represent a tool for objective assessment of celiac 13 . Cassinotti A , Birindelli S , Clerici M et al. HLA and autoimmune digestive disease when diagnosis is diffi cult. disease: a clinically oriented review for gastroenterologists . Am J Gastroen- WHAT IS NEW HERE terol 2009 ; 104 : 195 – 217 . 3 1 4 . F a i s S , M a i u r i L , P a l l o n e F et al. Gliadin induced changes in the expression The challenge in vitro of duodenal mucosa with gliadin is of MHC-class II antigens by human small intestinal epithelium. Organ useful and effi cient for establishing a gluten response and

culture studies with coeliac disease mucosa . Gut 1992 ; 33 : 472 – 5 . COLON/SMALL BOWEL allows celiac disease diagnosis in difﬁ cult cases. 15 . Tosco A , Maglio M , Paparo F et al. Immunoglobulin A anti-tissue trans- 3 The in vitro gliadin challenge with testing of mucosal glutaminase antibody deposits in the small intestinal mucosa of children expression of HLA-DR at 3 h of culture with gliadin in with no villous atrophy. J Pediatr Gastroenterol Nutr 2008 ; 47 : 293 – 8 . 16 . Salmi TT , Collin P , Korponay-Szabo IR et al. Endomysial antibody-negative comparison with blank is a simple procedure and coeliac disease: clinical characteristic and intestinal autoantibody deposits . might be less expensive than repeated endoscopies Gut 2006 ; 55 : 1746 – 53 . and biopsies. 1 7 . M a i u r i L , A u r i c c h i o S , C o l e t t a S et al. Blockage of T-cell costimulation 3 It should be performed in tertiary centers for the diagnosis inhibits T-cell action in celiac disease . Gastroenterology 1998 ; 115 : 564 – 72 . and therapy of celiac disease and should be strictly limited 1 8 . K a u k i n e n K , P e r ä a h o M , C o l l i n P et al. Small-bowel mucosal transglutami- to cases in which a GFD or non-concordant tests interfere nase 2-specifi c IgA deposits in coeliac disease without villous atrophy: a prospective and randomized clinical study . Scand J Gastroenterol with a correct diagnosis of celiac disease. 2005 ; 40 : 564 – 72 . 19 . Korponay-Szabo IR , Halttunen T , Szalai Z et al. In vivo targeting of intestinal and extraintestinal transglutaminase 2 by coeliac autoantibodies . Gut 2004 ; 53 : 641 – 8 . REFERENCES 20 . Bonamico M , Sabbatella L , Di Tola M et al. Antiendomysial antibody 1 . Green PH , Cellier C . Celiac disease . N Engl J Med 2007 ; 357 : 1731 – 43 . detection in biopsy culture allows avoidance of gluten challenge in celiac 2 . Di Sabatino A , Corazza GR . Coeliac disease . Lancet 2009 ; 373 : children. J Pediatr Gastroenterol Nutr 2005 ; 40 : 165 – 9 . 1480 – 93 . 2 1 . P i c a r e l l i A , M a i u r i L , F r a t e A et al. Production of antiendomysial antibod- 3 . O b e r h u b e r G , G r a n d i t s c h G , V o g e l s a n g H . Th e histopathology of coeliac ies aft er in-vitro gliadin challenge of small intestine biopsy samples from disease: time for a standardized report scheme for pathologists . Eur J Gas- patients with coeliac disease . Lancet 1996 ; 348 : 1065 – 7 . troenterol Hepatol 1999 ; 11 : 1185 – 94 . 2 2 . C a r r o c c i o A , I a c o n o G , D′ Amico D et al. Production of anti-endomysial 4 . Ciacci C , Cavallaro R , della Valle N et al. Th e use of serum tTG-ab assay in antibodies in cultured duodenal mucosa: usefulness in coeliac disease patients on gluten-free diet as a measure of dietetic compliance . Gastroen- diagnosis . Scand J Gastroenterol 2002 ; 37 : 32 – 8 . terology 2002 ; 122 : 588 . 23 . Maiuri L , Ciacci C , Auricchio S et al. Interleukin 15 mediates epithelial 5 . K a u k i n e n K , P a r t a n e n J , M ä k i M et al. HLA-DQ typing in the diagnosis of changes in celiac disease . Gastroenterology 2000 ; 119 : 996 – 1006 . celiac disease. Am J Gastroenterol 2002 ; 97 : 695 – 9 . 2 4 . B i a g i F , B i a n c h i P I , C a m p a n e l l a J et al. Th e prevalence and the causes of 6 . Rashtak S , Murray JA . Tailored testing for celiac disease. Ann Intern Med minimal intestinal lesions in patients complaining of symptoms suggestive 2007 ; 147 : 339 – 41 . of enteropathy: a follow-up study. J Clin Pathol 2008 ; 61 : 1116 – 8 . 7 . L u n d i n K E , S c o t t H , H a n s e n T et al. Gliadin-specifi c, HLA-DQ (alpha 2 5 . Wa h n s c h a ff e U , S c h u l z k e J D , Z e i t z M et al. Predictors of clinical response to 1*0501,beta 1*0201) restricted T cells isolated from the small intestinal gluten-free diet in patients diagnosed with diarrhea-predominant irritable mucosa of celiac disease patients . J Exp Med 1993 ; 178 : 187 – 96 . bowel syndrome . Clin Gastroenterol Hepatol 2007 ; 5 : 844 – 50 .