Oral pathology in untreated coelic disease



This article is corrected by:

  1. Errata: Corrigenda Volume 27, Issue 6, 528–529, Article first published online: 20 February 2008

Dr Prof. A. Carroccio, Internal Medicine, University Hospital of Palermo, via del Vespro 141, 90127 Palermo, Italy.
E-mail: acarroccio@hotmail.com


Background  Many coeliac disease patients with atypical symptoms remain undiagnosed.

Aim  To examine the frequency of oral lesions in coeliac disease patients and to assess their usefulness in making coeliac disease diagnosis.

Patients and methods  One hundred and ninety-seven coeliac disease patients and 413 controls were recruited and the oral examination was performed.

Results  Forty-six out of 197 coeliac disease patients (23%) were found to have enamel defects vs. 9% in controls (< 0.0001). Clinical delayed eruption was observed in 26% of the pediatric coeliac disease patients vs. 7% of the controls (< 0.0001). The prevalence of oral soft tissues lesions was 42% in the coeliac disease patients and 2% in controls (< 0.0001). Recurrent aphthous stomatitis disappeared in 89% of the patients after 1 year of gluten-free diet. Multi-logistic analysis selected the following variables as the most meaningful in coeliac disease patients: dental enamel defects (OR = 2.652 CI = 1.427–4.926) and soft tissue lesions (OR = 41.667, CI = 18.868–90.909). Artificial Neural Networks methodology showed that oral soft tissue lesions have sensitivity = 42%, specificity = 98% and test accuracy = 83% in coeliac disease diagnosis.

Conclusions  The overall prevalence of oral soft tissue lesions was higher in coeliac disease patients (42%) than in controls. However, the positive-predictive value of these lesions for coeliac disease diagnosis was low.


Coeliac disease (CD), one of the most common chronic diseases among Caucasians, is an autoimmune enteropathy triggered by the ingestion of gluten-containing grains in genetically predisposed subjects. During recent decades, the use of serological assays with a good diagnostic accuracy has shown that its frequency is much higher than previously considered, ranging between 1:85 and 1:300 both in Europe and in the USA.1–7 At the same time, it has become evident that the ‘typical’ clinical presentation of CD, i.e. the malabsorption syndrome, is nowadays less frequent and many CD patients show ‘atypical’– non-gastroenterological – symptoms or are asymptomatic8–12. This different clinical presentation makes diagnosis difficult; in fact, a serological screening study revealed that the ratio of known CD cases to undiagnosed CD cases was 1–7.13 Lack of diagnosis may have important consequences as dietary avoidance of gluten determines remission of the disease and avoids the onset of malignancy and thus decreases mortality in CD patients.14–16 Consequently, paediatricians, gastroenterologists and internists must look beyond the intestine to suspect a CD diagnosis. In fact, haematological, dermatological, neurological, obstetric, gynecological and the most proteiform clinical presentations of CD should be considered.17, 18 In this respect, the oral cavity, an apparatus which is very easy to examine, can be very interesting. In fact, mainly recurrent aphthous stomatitis (RAS) and dental enamel defects have been reported to be associated with CD.19–24

The primary aim of this study was to assess the frequency of oral diseases, including soft and hard tissue lesions in CD patients, in comparison with otherwise healthy controls; the secondary objective was to consider what contribution an oral clinical examination could give to help suspect a CD diagnosis.

Patients and methods

The total study population of this prospective study consisted of 610 subjects; of these, 197 were CD patients, recruited at the time of diagnosis. They were consecutively enrolled, between January 2004 and June 2005: the adult patients (90 cases: 65 F, 25 M, age range 18–75 years, median 31) in two centres – gastroenterology and internal medicine – of the University Hospital of Palermo and the children (107 cases: 59 F, 48 M, age range 2–17 years, median 9 years) in the Pediatric Gastroenterology Unit of the ‘Di Cristina Hospital’ in Palermo.

Coeliac disease patients aged under 2 years were excluded.

Four hundred and thirteen healthy subjects who were age/sex-matched (180 adults: 130 F, 50 M age range 19–77 years, median 32; 233 children: 120 F, 113 M, age range 2–17 years, median 8.5) and living in the same geographical area as that of the CD group were enrolled as controls. Paediatric controls were recruited (by simple randomization) at a day nursery, and at Primary and Secondary Schools during a health prevention programme for oral diseases; these subjects did not refer any diseases, had no family history of CD and showed normal growth (weight/height ratio between 25th and 75th centiles). Adult controls were recruited among otherwise healthy patients consecutively referred to the Dental Unit of the University of Palermo for third molar surgery; they were tested for serum anti-transglutaminase (anti-tTG) antibodies and were negative.

Coeliac disease diagnosis was based on the positivity of serum anti-tTG and/or anti-endomysium (EmA) antibodies, presence of clinical symptoms and positive histological evidence of villous atrophy with crypt hyperplasia and increase in intraepithelial lymphocytes on a gluten-containing diet, and the disappearance of the symptoms and normalization of serum anti-tTG and/or EmA on a gluten-free diet (GFD).25

The clinical manifestations of CD were classified as ‘typical’ when they included: chronic diarrhoea, failure to thrive, anorexia, abdominal distension and muscle wasting; other clinical manifestations were considered ‘atypical’.17 When CD diagnosis was made in subjects who were apparently asymptomatic, it was classified as ‘silent’.

Immediately after CD diagnosis, the patients underwent an intra-oral examination of the soft and hard tissues. All evaluations were performed independently by two of the authors (C. D. and D. L. C., who were trained in oral health survey) and then tested for concordance.

We focused on hard tissue lesions (i.e. dental enamel defects), soft tissue lesions (e.g. presence of RAS, aspecific atrophic glossitis and geographic tongue) and clinical delay of dental eruption.

The enamel defects affecting deciduous and permanent teeth were graded 0–IV according to Aine’s classification.26 As regard RAS, we included recurrent ulcerative lesions clinically observed by two of the authors (C. D. and D. L. C.) during the intra-oral examination. However, as clinical evidence of the lesions is not always found because healing occurred before the oral investigation, but patients could have had a past history of aphthous ulcers, we also considered ulcerative events, noted by parents or patients, or reported in hospital clinical records, with clinical features pathognomonic of RAS.

To evaluate delayed eruption in the paediatric patients, we used the conventional eruption tables for the Caucasian population27 and we considered delayed eruption as when the teeth were not in arch after their normal age of eruption, with a range of ±6 months.

In all individuals, dental hygiene was categorized into nominal variables using three values: 0 (poor), 1 (sufficient) and 2 (good).28

All CD patients with oral soft tissue lesions were re-evaluated 1 year after the beginning of GFD.

Finally, all the paediatric controls positive for oral hard and/or soft tissue lesions potentially associated with CD were tested for anti-tTG antibodies to exclude the disorder.

Informed consent was obtained for all participants in the study and the study was approved by the Ethics Committee of the University Hospital of Palermo.

Statistical analysis

Data were analysed by means of sas for Windows ver. 9.0, (SAS Institute Inc., Cary, NC, USA) and by means of statistica 6.0 (Statsoft Inc., Tulsa, USA). To measure the association level, crude odds ratio (OR) and the 95% corresponding test-based CI were calculated. Student’s t-test was used to calculate significant differences between cases and controls at baseline for normally distributed variables. The Mann–Whitney U-test was used to calculate differences between not-normally distributed variables (e.g. Aine’s scores).

The concordance rate of the oral evaluation between the observers was evaluated using Cohen’s kappa statistic, measuring agreement beyond that expected by chance (expressed as a coefficient ranging from 0 to 1.00).29

The relationship between CD and other variables was analysed within conditional multivariate frameworks.

A conditional logistic regression model was constructed stratified by age and dental hygiene; the same model permitted a stepwise selection procedure to obtain the most parsimonious model. The maximum likelihood ratio and adjusted OR were obtained using the iterative weighted least squares procedure. In all of the evaluations, P-values ≤0.05 were considered statistically significant.

Furthermore, in a blind manner, three Artificial Neural Network (ANNs) models were applied to the same data, as previously described.30 Fully-connected multilayer feedforward networks were used. The learning rule employed was the well-known back-error propagation, which adjusts the internal parameters of the networks over the repeated training cycles to reduce the overall error.31 The networks were validated with a new set of data, different from the training ones. The performance measured by mean squared error (MSE), accuracy, sensitivity and specificity values as well as the receiver operating characteristic (ROC) area, which can ascertain the degree of meaningful prediction,32, 33 was calculated for the significant associations. In this study, all variables were selected as input ones, except for CD (the output variable) for ANNs system analysis.


There was a substantial agreement (kappa = 0.85) between the observers for the main dental and oral lesions.

Table 1 summarizes the main findings of the oral examination in CD patients and controls.

Table 1.   Oral examination findings in coeliac disease (CD) patients and controls, grouped by adult and paediatric age
 Adult individualsPaediatric individuals
CD patients (n = 90) Controls (n = 180) P-valueUnadjusted OR (95% CI)CD patients (n = 107) Controls (n = 233) P-valueUnadjusted OR (95% CI)
Dental enamel defects (%)
 No69 (77)164 (92)0.0013.12 (1.53–6.33)82 (77)212 (91)0.00033.07 (1.63–5.80)
 Yes21 (23) 16 (8)25 (23)21 (9)
Soft tissue lesions (%)
 No56 (62)177(97)<0.000121.49 (8.02–57.59)59 (55)229 (98)<0.000146.57 (16.14–134)
 Yes34 (38)  5 (3)48 (45)4 (2)
Recurrent aphthous stomatitis (%)
 No71 (79)179 (99.5)<0.000147.36 (6.22–360.53)89 (83)231 (99)<0.000123.35 (5.31–102)
 Yes19 (21)  1 (0.5)18 (17)2 (1)

Forty-six out of 197 CD patients (23%) were found to have systematic and symmetrical enamel defects vs. a lower rate of 9% (37/413) in controls [< 0.0001; OR = 3.510 (95% CI = 2.135:5.770)]. The frequency of the enamel defects was very similar in the adult and paediatric CD patients. The severity of enamel defects in CD patients, evaluated according to Aine (17), was: grade I in 87%, grade II in 11% and grade IV in 2%. Figure 1a,b shows grades 1 and 4 lesions, respectively, observed in CD patients who were included in the study.

Figure 1.

 (a) Enamel defects of upper incisors (arrows indicate the color changes); (b) Enamel defects of upper incisors (structural changes); (c) major type of recurrent aphthous stomatitis and (d) numerous minor-type lesions of RAS.

Clinical delayed eruption was observed in 28 out of 107 paediatric CD patients (27%) vs. 16 out of 233 (7%) controls [< 0.0001; OR = 5 .932 (95% CI = 3.407:10.330)].

The overall prevalence of oral soft tissue lesions was 42% (82/197) in CD patients and 9/413 in controls (2%) [< 0.0001; OR = 22.257 (95% CI = 13.828:35.824)]; frequency was similar in adult and paediatric CD. RAS was found in 37/197 (19%) CD patients vs. 3/413 (1%) controls [< 0.0001; OR = 18.9505 (95% CI = 9.552:37.595)]. In CD patients, RAS was directly observed during the medical visit in 34 cases and simply recalled by the parents in three cases. Figure 1c,d shows an image of RAS observed in CD patients who were included in the study. The other soft tissue lesions detected were aspecific atrophic glossitis and geographic tongue. The first was found in 31/197 (16%) CD patients vs. 1/413 (0.2%) controls [P < 0.0001; OR = 22.464 (95% CI = 10.500:48.063)], while geographic tongue was noted in 14/197 (7%) CD patients vs. 5 out of 413 (1%) controls [P < 0.0001; OR = 7.0326 (95% CI = 2.650:18.666)]. CD patients showed a better dental hygiene status than controls [< 0.0001; OR = 4.848 (95% CI = 2.7027:8.695)].

None of paediatric controls with oral hard and/or soft tissue lesions presented positive serological markers for CD.

As regards the clinical manifestations of CD, ‘typical’ symptoms were more often observed in children (60/107 cases, 56%) than in adults (40/90 cases, 44%), whereas the frequency of silent cases was similar (7% in adults vs. 7% in children). However, oral hard or soft tissue lesions were observed with an almost identical frequency in patients with ‘typical’ and ‘atypical’ CD symptoms. Furthermore, as among patients without any signs and symptoms potentially related to CD, as diagnosed during familial CD screening, we found cases of oral soft tissue lesions, they should be considered as patients with ‘atypical’ and not with ‘silent’ CD. In CD patients, the frequency of the oral hard and/or soft tissue lesions did not differ according to the severity of intestinal mucosa damage or to the ideal body weight.

After a 1-year follow-up, 33 out of 37 CD patients (89%) with RAS at diagnosis referred that they had no longer suffered from RAS since beginning the GFD. The other four patients (11%) did not strictly adhere to GFD, as confirmed by persistently elevated serum anti-tTG antibodies, and did not report any improvement in RAS recurrence and number of ulcers per episode. Also, atrophic glossitis disappeared in all the patients who adhered to GFD. As regards geographic tongue, no cases were present in the sample followed longitudinally.

Conditional multi-logistic analysis in the stepwise procedure selected Oral Mucosa Lesions and Dental Enamel Hypoplasia as the most meaningful variables in CD patients (Table 2).

Table 2.   Conditional multi-logistic analysis in the stepwise procedure. Characteristics and risk factors were stratified by age and oral care
 Odds ratio95% CI
Gender (female vs. male)1.9801.253–3.130
Dental enamel hypoplasia2.6521.427–4.926
Oral soft tissue lesions41.66718.868–90.909

Artificial Neural Networks methodology consistently proved that CD was the most meaningful variable related to soft tissue lesions in the present dataset, with MSE equal to 0.321. Performance indexes showed the following values: accuracy = 83%; sensitivity = 42%, specificity = 98%. The ROC area was equal to 0.83.

On the basis of the 3% prevalence of CD recorded in our centres during the study period, and on the basis of the sensitivity and specificity shown by the ANN methodology, the positive and negative predictive values of the oral lesions were 39% and 99%, respectively. On the basis of the 1% CD prevalence in the general population, the positive and negative predictive values of the oral lesions in CD diagnosis were 17% and 99%, respectively.


Recent epidemiology data have shown a prevalence of CD approaching 1% in the general population.34–36 However, there has been a noticeable change in the clinical presentation of CD, as almost 50% of the patients with newly diagnosed CD do not present with gastrointestinal (GI) symptoms37, 38 thus making diagnosis difficult. This is a paramount aspect as a GFD in CD patients is known to prevent many of the extra-intestinal symptoms, such as osteoporosis,39 recurrent abortions,40 and above all, it protects against the development of cancer.41 Thus, to identify the greatest number of ‘atypical’ or ‘silent’ CD patients and prevent complications, clinicians must investigate ‘at-risk subjects’, e.g. those with chronic anaemia,42, 43 hyper-transaminasemia or hyperamylasemia of unknown origin,44, 45 osteoporosis,39 autoimmune thyroid disorders.46 Furthermore, it is known that CD is an auto-immune disease resulting from an inappropriate T cell-mediate immune response against ingested gluten.47 Although the proximal part of the intestinal mucosa represents the main site of the gut involved in CD, it has been demonstrated that gluten-driven T-cell activation is not restricted to the small intestine, but is present in the whole GI tract. The mouth, the first part of the GI system, represents a site very easy to detect and an oral examination could give a useful diagnostic contribution as lesions of the hard48, 49 and soft tissues50, 51 have been reported in CD.

Ours is the largest uni-centre study to have investigated the risk for CD patients of suffering from dental or oral mucosa lesions and shown the sensitivity and specificity of oral soft tissue lesions in suggesting a CD diagnosis.

As regards the hard tissues, we found systematic and symmetrical enamel defects in 23% of CD patients, with an OR >3 in comparison with the controls. The enamel defects resulted in a dental deformity, which can be easily recognized (see Figure 1), although low-grade lesions must also be accurately investigated. Other studies have reported a wide range of frequencies of enamel defects in CD patients,26, 48, 52–59, but our data are in agreement with other studies performed in Italy, and the differences in frequency probably depend on environmental, dietetic and, above all, genetic factors.54 The same hypothesis could be made for the severity of enamel defects which appeared less severe in our study (87% of the patients had a grade I lesion) than in other studies.48

The aetiopathogenesis of these defects in CD patients still remains unclear. As the crowns of deciduous and permanent teeth develop from 4 to 5 months of intra-uterine life to the 7th year of life except for the third molar, nutritional, immunological or genetic factors (association with the HLA DR3 allele) have been hypothesized as causing developmental defects in the enamel.53, 60 Hypocalcaemia caused by malabsorption during dental development has been considered to be implicated in enamel hypoplasia. Regarding the nutritional dynamics, despite some doubts raised by Maki et al.,60 it has been hypothesized that a gluten-induced immunological process could occur between 6 months and 7 years in the enamel-producing organ, resulting in defective enamel formation. Finally, these dental anomalies have been found to be significantly related to HLA antigen DR3.53, 56, 60 Furthermore, an Italian study,53 reporting a frequency of 29% in coeliac patients vs. 15% in controls, did not report any statistically significant differences in calcium concentrations, but a coincidence in 77% of CD and enamel defects in DR3-positive subjects. Another finding on dental hard tissues was the significantly higher frequency of delayed eruption, observed in 27% of CD children. Only few papers in the literature have dealt with this issue,49, 61–63 and they are in agreement with our data. The delayed dental eruption could be seen as a possible sign of malnutrition (such as is delayed puberty) and advises for serological CD screening.

However, the most important finding of the present study is related to the oral soft tissue investigation: in fact, mucosa lesions were found in 42% of CD patients with an OR of 22 vs. controls. Within this group of lesions, RAS was found in 37/197 (19%) CD patients with an OR of 19 vs. controls. RAS frequency in CD observed in our study is in agreement with that reported by other authors.64, 65 It is also very interesting that almost all (89%) the CD patients with RAS no longer suffered after beginning GFD and the lack of healing in the remaining patients was probably linked to the lack of adherence to GFD. Consequently, RAS persistence in CD patients could cautiously be considered a marker of lack of GFD adherence. This hypothesis is supported by a study which reported that RAS and intestinal histological alterations relapsed after gluten challenge.66

Although several studies have reported the presence of oral mucosa lesions in CD, our study reports the first evaluation of the risk of such lesions in CD patients using univariate, multi-logistic regression, ANN sensitivity and specificity testing.

However, despite a good test accuracy, the presence of oral lesions showed a low positive-predictive value of 16%, giving in the general population a 1% prevalence of CD.

In conclusion, our study showed a higher frequency of oral alterations in CD patients in comparison with healthy controls. However, the presence of these lesions had a low positive-predictive value in CD diagnosis.

Authors’ contributions

All the authors participated in drafting the article and/or revising it critically and gave the final approval of the version to be published. Furthermore, in particular, they worked as follows: Giuseppina Campisi, Chief Investigator of the Dental Care Unit, did the conception and design of the study and interpretation of the data. Chiara Di Liberto, Investigator of the Dental care Unit, was in charge of the acquisition of data. Giuseppe Iacono, Chief investigator of the Paediatric Gastroenterology Unit, was in charge of the acquisition of data. Domenico Compilato, Investigator of the Dental care Unit, was in charge of the acquisition and interpretation of data. Lidia Di Prima and Carmela Sferrazza, Investigators of the Internal Medicine Unit, were in charge of the acquisition of data. Francesco Calvino was responsible for the statistical analysis, analysis and interpretation of the data. Vito Di Marco, Investigator of the Gastroenterology Unit, was in charge of the acquisition of data. Lorenzo Lo Muzio did the analysis and interpretation of data. Calogero Scalici, Investigator of the Paediatric Gastroenterology Unit, was in charge of the acquisition of data. Craxì Antonio, Chief Investigator of the Gastroenterology Unit, did the conception and design of the study. Carroccio Antonio, Chief Investigator of the Internal Medicine Unit, did the conception and design of the study and interpretation of data.


We thank Ms Carole Greenall for her precious help in revising the English language used. Declaration of personal and funding interests: This study was supported by a grant of the MIUR (PRIN 2005, number 2005069443) to Prof. Carroccio and Prof. Campisi, and by a grant of the University of Palermo (project ex 60%, 2004) to Prof. Carroccio. Prof. Giuseppina Campisi is the guarantor of the paper. All the Authors have declared that none of them has potential competing interests and they have not received other financial support for the study.