Grown-up coeliac children: the effects of only a few years on a gluten-free diet in childhood

Authors


Dr C. Ciacci, Gastroenterologia, via Pansini 5, 80131 Napoli, Italy.
E-mail: ciacci@unina.it

Summary

Aim : To evaluate clinical and psychological status of adults with childhood diagnosis of coeliac disease who were re-exposed to gluten after only a few years and now on a gluten-containing diet, compared with adults with recent diagnosis of coeliac disease, and adults who remained on gluten-free diet after childhood diagnosis.

Methods : A total of 195 adults with a biopsy suggestive of coeliac disease in childhood, who either had adhered to a gluten-free diet for at least 1 year after diagnosis and now are either on gluten-free diet (n = 110) or on gluten-containing diet (n = 85), and adults with newly diagnosed coeliac disease (n = 165) underwent a medical check-up.

Results : Body mass index and main laboratory indices were statistically different among groups (lowest in never on gluten-free diet, highest in gluten-free diet). The lowest average levels of bone mineral density were found among never on gluten-free diet patients. Prevalence of autoimmune disorders was increased in never on gluten-free diet when compared with the transient gluten-free diet and gluten-free diet groups. Histology revealed villous subatrophy in all patients of never on gluten-free diet group, in 39 of 110 patients of gluten-free diet and in 84 of 85 of transient gluten-free diet groups. Herpetiform dermatitis was found in three patients of gluten-free diet, three of transient gluten-free diet and three of never on gluten-free diet. Dental enamel defects were found in 15 patients of transient gluten-free diet, 43 of never on gluten-free diet and in zero of the gluten-free diet group. Pregnancy outcome was not significantly different between the two groups, but neonatal weight was lower and breast feeding was shorter in the never on gluten-free diet group. Sexual habits, alcohol intake and cigarette smoking were significantly different in the never on gluten-free diet group when compared with the other two groups.

Conclusion : Gluten withdrawal in childhood partly protects coeliac adults from clinical and behavioural effects of gluten sensitivity.

Introduction

Coeliac disease is a condition in which, in genetically predisposed people, there is a reaction to gluten, a protein in some cereal grains.1, 2 This reaction causes the formation of autoantibodies and the destruction of the villi in the small intestine, resulting in malabsorption of nutrients. Lifelong gluten withdrawal from diet is now considered as the sole main therapy.

Once thought to be a childhood disease that would be outgrown, diagnosis of coeliac disease was mainly made on a clinical and histological basis.3, 4 In Europe, after a variable period of being on a gluten-free diet (GFD), children were re-exposed to gluten.5 In the case of absence of clinical signs, patients were often left on a gluten-containing diet. Recent evidence indicates that it is common for the symptoms of coeliac disease to disappear during late childhood or adolescence, implying an apparent cure.6 Unfortunately, overall health still suffers: later in life these coeliac patients may find that there is considerable damage to the small intestine7, 8 and for years might have deprived themselves of important nutrients, such as calcium.9,10 In more recent years, better knowledge of the disease and its frequency among adults11 encouraged many of them to seek treatment for coeliac disease.

The aim of the present study was to evaluate nutritional indices, bone mineralization, pregnancy outcome and some psychological features of adults with childhood diagnosis and diet therapy for coeliac disease in comparison with matched adults with newly diagnosed coeliac disease.

Patients and methods

The study was designed on a prospective basis. A total of 215 adult patients with diagnosis of coeliac disease in childhood and 165 sex- and age-matched adults with newly diagnosed coeliac disease were consecutively examined in the tertiary outpatient clinic for malabsorptive diseases of the Federico II University of Napoli, Italy. Patients with previous diagnosis of coeliac disease were enrolled if they had, in childhood, a biopsy suggestive of coeliac disease and if they had been on a gluten-containing diet for at least 1 year after diagnosis. The medical check-up included: anthropometrics, physical examination, laboratory tests [including checks for gluten-related specific autoantibodies anti-endomysium (EMA) and anti-tissue-transglutaminase antibodies (a-tTG) both from Eurospital, Trieste, Italy], abdominal ultrasound, upper endoscopy and duodenal (two specimens) and jejunal biopsies (three specimens), bone densitometry (BMD) measured by dual energy X-ray absorptiometry (Hologic, Waltham, MA, USA), and every other specific investigation that the single clinical patient condition required (i.e. dermatology, gynaecologist consultation, etc.).

Patients were divided into three groups: the GFD group consisted of 110 patients with proved coeliac disease diagnosis in childhood who have been on GFD; the transient gluten-free diet (TGF; on transient GFD = TGF) group was made up of 85 patients with proven coeliac disease diagnosis in childhood, who after a variable time period (at least 1 year) were re-exposed to gluten and are now on gluten-containing diet; the third group was made up of 165 sex- and age-matched patients with new coeliac disease diagnosis in adulthood examined before starting the GFD (>18 years old, never on GFD = NGF). All participants underwent a dietetic consultation. A self-rating scale was administered to evaluate coeliac disease patients’ GFD ‘level of knowledge’ of the disease. They were asked to answer, in a multiple-choice paradigm, a standardized questionnaire about coeliac disease. This scale, the Knowledge Rating Scale (KRS),12 rates answers from 0 to 25, where 25 corresponds to the best knowledge. The patients on GFD also were asked to complete a visual analogue scale, the Dietetic Restriction Compliance (DRC),12 which consisted of an unmarked 10-cm line with the anchor sentence ‘I never comply with my diet’ and ‘I always comply with my diet’ at each end. The DRC rates the answers from 0 to 10, where 10 corresponds to a complete self-reported adherence to dietetic restriction.

All participants underwent psychological consultation with a single psychologist (MS). During the interview some questionnaires were administered: standardized questionnaire on depression, the modified Zung Self-rating Depression Scale (M-SDS)12 and on sexual habits.13 The questionnaires were administered after reassuring the patients of the anonymity of the entire procedure of the data analysis.

In 11 patients with a previous diagnosis of coeliac disease and now on a normal diet, EMA, IgA and a-tTG search were negative and coeliac disease diagnosis was not confirmed. Nine of EMA-positive patients (from TGF group) refused to undergo an endoscopy for fear of such a procedure and did not participate in the study. The study cohort was finally made up of 195 adult patients with diagnosis of coeliac disease in childhood and 165 adults with coeliac disease diagnosis in adulthood, with complete data. Figure 1 shows the algorithm of the recruitment of the study cohort.

Figure 1.

The study cohort and how the three groups of coeliac patients that entered in the analysis were made-up.

Histological examination of duodenal biopsies was performed according to Oberhuber et al. modification14 of Marsh criteria.15 All biopsies were oriented on paper at the moment of formalin solution fixation.

Computation was carried out by the spss software package for Windows (release 11.5.1; SPSS Inc., Chicago, IL, USA). Student's t-test for unpaired observations, anova and post hoc Bonferroni test, univariate and multivariate regression analyses, and analysis of 95% confidence intervals (CI) were used.

Results

Anthropometry, clinical and laboratory data

Clinical records, prevalence and type of symptoms at diagnosis, and histology at diagnosis of all patients with childhood diagnosis of coeliac disease indicated that there were no statistically significant differences between patients who stayed on GFD and those who were re-exposed to gluten after a variable period of time (data not shown). The main difference noted in the paediatricians’ decision was the hospital in which the diagnosis was made (university or city hospitals) and the general practitioner's advice. In seven patients of GFD group who were re-exposed to gluten, coeliac disease symptoms relapsed and they were put on a GFD again.

Table 1 shows gender, age at diagnosis and age at examination, anthropometry and main selected laboratory tests of the three groups under study. Most nutritional indices were significantly lower in the groups on normal diet (TGF and NGF) when compared with GFD group, and statistically significant linearity was found for most values among groups. Forty-five patients of GFD group were EMA-positive or a-tTG-positive and/or showed gluten-related damage at intestinal biopsy histology. Thus, only 65 patients were strictly compliant to GFD. Before proceeding with the complete data analysis, we compared the 45 non-compliant with the 65 strictly compliant patients. All the nutritional indices were not statistically different between the two subgroups (Table 2). Therefore, for the rest of the analyses the two subgroups were analysed together.

Table 1.  Gender, age at diagnosis, age at visit, duration of GFD, BMI and selected laboratory indices of the cohort under examination
 GFD (n = 110)TGF (n = 85)NGF (n = 165)P-value (anova)
  1. Test for linearity: BMI: F = 45.360, P = 0.000; ferritin: F = 3.310, P = 0.071; HGB: F = 6.872, P = 0.009; total protein: F = 54.783, P = 0.000; cholesterol: F = 25.590, P = 0.000; AST: F = 1.874, P = 0.173.

  2. * Normal values <5, doubt 5–7, pathological >7.

  3. GFD, gluten-free diet; BMI, body mass index; CD, coeliac disease; EMA, anti-endomysium; a-tTg, anti-tissue-transglutaminase; TGF, transient gluten-free diet; NGF, never on gluten-free diet.

Gender (M/F)37/7328/5751/114N.S.
Age at visit (years)21.9 ± 3.723.7 ± 4.323.1 ± 3.20.060
Age at CD diagnosis (years)0.9 ± 1.53.60 ± 1.723.1 ± 3.2
Years of GFD (years)15.7 ± 6.02.2 ± 0.90.001
Body weight (kg)58.8 ± 10.657.8 ± 9.956.3 ± 9.8N.S.
Height (cm)165 ± 4.2164 ± 8.6163 ± 9.7N.S.
BMI (kg/h2)21.7 ± 2.720.3 ± 3.018.96 ± 3.480.001
EMA-positive (n)2185165
a-tTg (AU)*10.2 ± 13.420.4 ± 9.928.3 ± 8.40.090
Serum ferritin (μg/mL)78.9 ± 37.057.2 ± 37.523.4 ± 47.80.001
Haemoglobin (g/dL)13.4 ± 1.412.3 ± 2.012.2 ± 3.20.002
Serum total protein (g/dL)7.2 ± 0.76.9 ± 0.86.8 ± 1.20.004
Serum albumin (g/dL)4.3 ± 0.74.2 ± 0.74.1 ± 0.60.190
Plasma cholesterol (mg/dL)150.2 ± 29.4140.9 ± 27.0138.5 ± 13.60.029
Serum alanine aminotransferase (U/L)22.2 ± 8.428.2 ± 20.744.4 ± 22.70.050
Table 2.  Gender, age at visit, BMI and selected laboratory indices of two subgroups of GFD patients divided in those on strict GFD vs. those on a non-compliant GFD
 Strict gluten- free diet (n = 65)Non-compliant gluten-free diet (n = 45)P-value
  1. GFD, gluten-free diet; BMI, body mass index.

Gender (M/F)20/4517/28
Age at visit (years)21.6 ± 3.422.7 ± 3.3N.S.
BMI (kg/h2)21.4 ± 1.720.8 ± 2.0N.S.
Serum ferritin (μg/mL)36.05 ± 38.333.78 ± 31.3N.S.
HGB (g/dL)13.3 ± 1.2512.2 ± 1.13N.S.
Serum total protein (g/dL)7.23 ± 0.707.18 ± 0.81N.S.
Serum albumin (g/dL)4.31 ± 0.744.25 ± 0.88N.S.
Plasma cholesterol (mg/dL)153.0 ± 31.3146.5 ± 35.2N.S.
Serum AST (U/L)21.3 ± 12.224.04 ± 11.9N.S.

Autoimmune thyroid disease was found in 14 women (two from GFD, two from TGF and eight from the NGF group), insulin-dependent diabetes mellitus (IDDM), already known, in two patients on GFD and one in the NGF group. Polyarthritis was reported by one patient in the TGF group and by six patients from the NGF group. Primary biliary cirrhosis was found in one patient from the NGF group. Prevalence of autoimmune disease was statistically different among the three groups (F = 11.791; P = 0.0001) and post hoc analysis showed that the main difference was due to the comparison from GFD and the NGF group (P = 0.001). Herpetiform dermatitis was found in nine patients (three from GFD, three from TGF and three from the NGF group). Dental enamel defects (their presence was detected by the doctors running the ambulatory, pictures were taken and then shown to a dentist) in at least one tooth were found in 15 patients on TGF, 43 from the NGF group and none in the GFD group. Prevalence of main gastrointestinal symptoms (dyspepsia, diarrhoea, abdominal pain, constipation) at the visit was not significantly different between the GFD and TGF groups (22% and 23% respectively) and present in 59 of 165 (35.7%) of patients in the NGF group (anova among groups: F = 10.965; P = 0.001). In 45% patients from GFD group reporting gastrointestinal symptoms both intestinal biopsy and laboratory test excluded gluten damage and the diagnosis was made up of functional disorders. Therapy for epilepsy was reported in three cases (two from GFD, one from TGF group), chronic headache in seven of GFD and in five from the NGF group. One of the patients, 19 years, non-compliant to GFD since childhood, died during examination because of intestinal lymphoma.

Pregnancy outcome

Table 3 shows selected data of the women participating in the study. Age of menarche was similar in the three groups. One woman from TGF and three from the NGF groups were undergoing hormone therapy for infertility. Number of pregnancies and offspring tended to be greater in NGF and TGF women than in those on GFD but it also must be noted that women in the TGF group tended to be older than those in NGF group. No differences were noted for global pregnancy outcome between the two groups, although the number of threatened abortion was significantly greater in TGF women compared with GFD women. Neonatal birth weight was significantly lower and the duration of breast feeding was shorter in the NGF and TGF women compared with GFD women. None of the examined women was in menopause.

Table 3.  Analysis of data of age of menarche and pregnancy outcome in women of GFD, TGF and NGF groups
 GFD (n = 73)TGF (n = 57)NGF (n = 78)P-value (anova)
  1. Values represent mean ± s.d. and absolute number in brackets.

  2. Test for linearity: months of breast feeding: F = 19.639, P = 0.0001; offspring birth weight: F = 23.423, P = 0.0001.

  3. GFD, gluten-free diet; TGF, transient gluten-free diet; NGF, never on gluten-free diet.

Age at menarche (years)12.6 ± 1.413.1 ± 1.712.5 ± 1.40.091
Pregnancies (n)0.22 ± 0.7 (11)0.47 ± 0.8 (16)0.5 ± 1.0 (32)0.074
Offspring (n)0.11 ± 0.4 (6)0.35 ± 0.7 (13)0.44 ± 0.8 (27)0.006
Abortion (n)0.11 ± 0.3 (3)0.10 ± 0.5 (4)0.09 ± 0.4 (6)0.956
Abortion-to-pregnancy ratio0.34 ± 0.40.22 ± 0.40.13 ± 0.30.360
Threatened abortion (n)0.0 ± 0.0 (0)0.9 ± 0.3 (5)0.0 ± 0.00.020
Preterm delivery (n)0.03 ± 0.27 (1)0.04 ± 0.20 (1)0.09 ± 0.40.870
Offspring birth weight (g)3290.2 ± 220.22892.3 ± 169.82478.0 ± 517.20.001
Months of breast feeding4.50 ± 2.12.56 ± 1.31.77 ± 0.50.001

Dietetic interview

Dietetic interview and visual analogue DRC scale revealed that 29 of 110 patients who were on GFD knew that they consumed gluten on a daily basis. KRS scale showed that 28 of 110 patients on GFD were unaware of gluten content of food that they consumed on a regular basis. In total, 34.5% of GFD patients were badly compliant to GFD. Patients from the TGF group, although aware of the fact that they were gluten intolerant in childhood did not report any significant reduction of gluten intake in comparison with the NGF group.

Intestinal biopsy

Histology revealed increased number of intraepithelial lymphocytes (>40 cells/100 enterocytes) and patchy subtotal villous atrophy in 12 patients and subtotal villous atrophy in 27 of 110 patients on GFD and in 84 of the TGF group. Normal histology was found in one EMA-positive patient of the TGF group. Histology of all biopsies from NGF patients showed subtotal villous atrophy.

Bone density

The BMD at femur was not significantly different among GFD (0.70 ± 0.02 g/cm2, mean ± s.d.) and TGF patients (0.71 ± 0.2 g/cm2). A significant decrease of BMD was found in the NGF group (0.64 ± 0.76 g/cm2). anova of BMD among the three groups showed a statistical significant difference (F = 23.798; P < 0.0001). Test for linearity showed a linear decrease of BMD among TGF, GFD and NGF patients (F = 45.360; P = 0.0001). Bonferroni's post hoc analysis indicated that the main difference accounted for the NGF group compared with both GFD and TGF groups (P = 0.0001). Linear regression analysis showed that BMD at femur is not related to the main nutritional indices and, in a separate analysis for patients on GFD, showed significant correlation with compliance score (Table 4).

Table 4.  Regression analysis: bone mineral density by nutritional indices and by the score calculated from the dietetic compliance analogue scale (the latter only in patients on GFD)
 rF95% CIP-value
  1. GFD, gluten-free diet; BMI, body mass index; CI, confidence interval.

BMI0.15215.300.582–0.6820.040
Serum total proteins0.1310.700.638–0.6880.053
Serum albumin0.8601.660.664–0.7570.339
Haemoglobin0.5700.920.620–0.6980.199
Plasma cholesterol0.1271.350.643–0.6840.065
Compliance score (n = 110)0.1964.040.622–0.6420.046

Psychological interview

All questionnaires were compared between groups with anova and post hoc test using M-SDS score as a dependent variable. M-SDS score data show that 22% of coeliac disease patients scored above the normal range at a P-value of 0.10 cut-off level. No statistical difference was found among GFD, NGF and TGF patients (F = 0.034; P = N.S.). This finding means that more than 20% of coeliac patients showed clinical signs of depression. Multiple regression analysis (dependent variable M-SDS score, independent variables age, gender, occupation, years of education, presence or not of gastrointestinal symptoms) showed that also demographic variables and symptoms did not influence depression level as indicated by the M-SDS score. Compliance to GFD (as indicated by DRC score) was strongly related to depression. In fact, the analysis of M-SDS scores limited to GFD patients showed that better the compliance, worse the depression (F = 36.063; P = 0.001; CI: 20.21 ± −2.543; upper 29.26 ± −1.279), independent from the duration in years of GFD (F = 0.136; P = N.S.).

The questionnaire on sexual habits was made up of two parts. Table 5 shows the results of the part dealing with indices of sexual behaviour in coeliac disease patients. Because of the young age of most participants and their religious belief, a number were not yet sexually active. Although not statistically significant, a greater number of patients of the NGF group were sexually active when compared with the other groups; the comparison among groups shows that the reported number of intercourse per month was statistically lower in NGF group. Post hoc analysis indicated that the main difference was accounted in the NGF group compared with both GFD and TGF groups. The second part of the questionnaire included questions on gender, age, menopause, use of oral contraceptives or other contraceptive devices, smoking habits (and if yes, number of cigarette/day), alcohol intake (if yes, g/day), education, job, marital status and religion. Selected items of the second part of the questionnaire are given in Table 6. Smoking and alcohol intake were statistically significantly more frequent in GFD patients compared with TGF and NGF patients.

Table 5.  Percentage values of selected indices of sexual behaviour in adult CD patients
 GFDTGFNGF
  1. *P = 0.001 (anova).

  2. CD, Crohn's disease; GFD, gluten-free diet; TGF, transient gluten-free diet; NGF, never on gluten-free diet.

Sexually active [n (%)]69 (62)65 (76)115 (69)
Frequency of intercourse/ month (mean ± s.d.)6.9 ± 1.98.7 ± 1.25.8 ± 1.1*
With low interest for sex (%)13.111.318.2
With difficulty in attaining orgasm (%)2.92.42.4
With painful intercourse (%)00.50.9
Satisfied with their sexual life (%)728971
Table 6.  Years of education, marital status, smoking and drinking habits, and occupation of CD patients
 GFDTGFNGF
  1. F = 5.019, P = 0.026 (anova).

  2. ** F = 7.760, P = 0.006 (anova).

  3. CD, coeliac disease; GFD, gluten-free diet; TGF, transient gluten-free diet; NGF, never on gluten-free diet.

Education (years)10.6 ± 2.5111.2 ± 3.911.8 ± 3.6
Marital status (n)
 Single626250
 Married311946
 Divorced27634
Smokers (n)342232
 Number of cigarette (day)9.6 ± 2.18.4 ± 1.27.8 ± 1.1*
Alcohol consumers (n)463024
 Alcohol intake (g/day)16.4 ± 2.110.3 ± 1.68.3 ± 1.6**
Occupation (n)
 Student351857
 House wife292336
 Blue collar151216
 White collar131637
 Professional456
 Unemployed131113

The interview revealed that none of the coeliac disease patients was habitual or recreational drug users.

Discussion

The present study describes a cohort of young adults with coeliac disease in whom diagnosis and subsequent therapy were made in childhood or adulthood. In the first case, as often happened in the past,3, 4 some patients were left on GFD and some others were re-exposed to gluten.

The main finding of the present study is that patients with childhood diagnosis and diet therapy who were on a transient GFD for few years and now on a gluten-containing diet show a clinical picture of coeliac disease less severe than that observed in patients with diagnosis in adulthood before starting the GFD. The finding allows the hypothesis that even a short period of GFD in childhood exerts a sort of protective effect from gluten-induced diseases.

The first observation against this hypothesis is that the TGF group is probably made up of persons with lower sensitivity to gluten compared with patients in the GFD group. In fact, in a number of patients in the GFD group re-exposed to gluten, the disease relapsed and GFD was again re-introduced, while in the TGF group no relapses were reported. This observation leads to the conclusion that some persons, after a time period of GFD, may develop a sort of resistance to the gluten toxic effect. Other clinical studies explore this possibility. Bardella et al.16 showed that about 20% of patients with proven herpetiform dermatitis experienced no clinical or histological relapses after a mean period of 15 years of gluten challenge after a short period of GFD in childhood. The immunological reason of such ‘resistance’ to gluten damage is still unknown. One of the possible mechanisms underlying resistance is represented by that of Zonula occludens toxin (Zot),17 a new toxin which is able to induce systemic tolerance to gliadin in animal experimental model.18 In this model, the reduced production of interferon-γ interleukin (IL)-2 upon administration of gliadin alone was reverted in mice receiving gliadin together with Zot. The gliadin-related down-regulation of Th1-like cytokines in this model was found to be associated with the suppression of the T-cell proliferation, while such a suppression was completely reverted by the presence of Zot co-administration.18 Zot and other molecules may have the ability to interfere with the suppression of specific cell-mediated immunity.

One-third of patients from the GFD group showed histology compatible with some gluten-induced damage (this finding confirmed by specific serum antibodies and dietetic interview and DRC scores). They might then have clinical and laboratory features closer to the TGF group than to healthy controls for their exposure to gluten. The finding is expected as it is known that in adults, compliance to GFD is often quite low.19 In our GFD group, prevalence of autoimmune diseases was similar to that of age- and sex-matched general population controls.20 Moreover, a separate analysis demonstrated that most nutritional indices of GFD patients who consumed gluten were not significantly different from those of the subgroup of GFD patients who were on a strict GFD. This finding suggests that, overall, the health status of badly compliant coeliac adult may be good anyway. The possibility of a safe threshold for gluten intake is to be considered although individual sensitivity is the major disadvantage in planning a study that can fully answer this question. The other possible bias in grouping the patients for the purposes of the study is that patients of the TGF group may have spontaneously reduced their gluten intake during the years of gluten exposure. The dietetic interview ruled out this possibility.

Our data of prevalence of autoimmune disease (similar in GFD and TGF groups) are only in apparent contrast with data from Ventura et al.21 who examined the relationship between the prevalence of autoimmune disorders in coeliac disease teenagers and the duration of exposure to gluten and concluded that the prevalence of autoimmune disorders in coeliac disease is related to the duration of exposure to gluten. A multicentre national study on adults showed that the prevalence of autoimmune diseases in patients with a late coeliac disease diagnosis does not correlate with duration of gluten intake.22 In the present cohort of young adults with coeliac disease diagnosed in adulthood (NGF group), prevalence of autoimmune disease is similar to that expected22 and similar to that reported by Ventura et al.21 and significantly greater than that found in GFD and TGF groups. The novelty of our finding is that for the prevalence of autoimmune disease, TGF patients are in between the two positions as it is possible that also for autoimmune disease a period of gluten withdrawal from diet early in the life of a coeliac child induces some protection.

Two studies underline the fact that early gluten exposure in infancy can be a determinant for autoimmunity. In the first study, it is demonstrated that food supplementation with gluten-containing food before the age of 3 months was associated with significantly increased pancreatic islet autoantibody risk without increase of risk of developing coeliac disease.23 In the second study, elevated mortality risk was found in coeliac persons mainly for associated immune disorders, whereas mortality risk in patients with coeliac disease diagnosis before the age of 2 was lower by 60% compared by the same age group of the general population.24 Therefore, coeliac disease diagnosis and early gluten withdrawal may have a protective effect against gluten-related immune disorders principally in first childhood.

Bone mineral density (BMD), although similar in GFD and TGF groups, show a linear decrease among groups from those on diet to the NGF group. This finding is in accordance with many reports of literature.24–26 Moreover, in GFD patients, BMD shows a significant direct correlation with self-reported compliance to GFD, reinforcing the idea of the importance of a strict GFD.27, 28

A novel finding to be highlighted is the increased dental enamel defects detected in TGF and NGF patients compared with GFD patients. The enamel abnormalities are easy to detect and our finding should encourage general practitioners and paediatricians to check for the specific signs and ask for dentist consultation or, in case of positive clinical history, proceed with specific autoantibody testing.

Pregnancy outcome is not statistically different among groups although neonatal birth weight and duration of breast feeding show a linear decrease from GFD to TGF and NGF group, indicating a strong effect of undiagnosed coeliac disease. Post hoc analysis indicated that the main differences accounted from comparison between GFD and NGF groups. The data seems to be in contrast with previous observations of increased abortion in untreated adult coeliac women,29, 30 but in our cohort the young age of participants may play a role in explaining the discrepancy. A different picture is given by the analysis of psychological status of the cohort under study.

The M-SDS indicates that depression is common also in young adults with coeliac disease 12, 31 regardless of the diet the patients were on. The finding is in accordance with the idea that depression is a feature of coeliac disease.12 The data from the questionnaire of sexual habits, instead, indicate that the NGF group reports reduced number of sexual intercourses and decreased percentage of people satisfied with their sexual life. The young age of most participants in GFD group limits the value of the finding, although it is in accordance with previous observations.13

Finally, the second part of the questionnaire on sexual habits points to the fact that there is a significant linear decrease of number of cigarettes smoked per day and alcohol intake from the GFD group to the NGF group. It can be speculated that NGF patients have in general poor health status that limits smoking habits32 and alcohol intake.

The psychological attitude of young adults with coeliac disease before dietetic treatment can be described as a low-energy mood that interfere with the general perception of quality of life. Few years of GFD in childhood from this point of view also seem to exert a beneficial effect, as the TGF group seem to do better than the NGF group who were never on GFD.

In conclusion, the main result of this study is that a short period of GFD during childhood or even a low compliance to diet for intentional or unintentional habitual gluten intake might have a long-term protective effect from gluten-related diseases.

Many coeliac patients pay a psychological price for the obligation imposed by dietary restriction33, 34 and are constantly reproached by doctors for their dietetic mistakes. Our data should not be misinterpreted: we still recommend strict adherence to GFD. Nevertheless, the findings of this study also raise the provocative question as to whether it is possible to discriminate among coeliac patient subgroups with low gluten sensitivity who still have good health status even by consuming small amounts of gluten. The possibility exists that this subgroup could be less restrictive with their GFD without excessive damage.

Acknowledgements

The authors are grateful to Ms Wendy Parsons for revising the manuscript. They are also grateful to Ms Carla Perrot, Lina Mautone and Patrizia Ferro for their nursing assistance to coeliac patients. Dr Siniscalchi's work was in part supported by Schär.

Ancillary