Digestive Diseases and Sciences

, 54:2175

Complications in Celiac Disease Under Gluten-Free Diet

Authors

    • Digestive Endoscopy Unit“Lorenzo Bonomo” Hospital
  • Walter Elisei
    • Department of Internal Medicine, Division of Gastroenterology“Cristo Re” Hospital
  • Gian Marco Giorgetti
    • Clinical Nutrition Unit“S. Eugenio” Hospital
  • Giovanni Brandimarte
    • Department of Internal Medicine, Division of Gastroenterology“Cristo Re” Hospital
  • Fabio Aiello
    • Department of Statistic and Mathematic SciencesUniversity of Palermo
Original Article

DOI: 10.1007/s10620-008-0595-1

Cite this article as:
Tursi, A., Elisei, W., Giorgetti, G.M. et al. Dig Dis Sci (2009) 54: 2175. doi:10.1007/s10620-008-0595-1

Abstract

We assessed the onset of malignant and nonmalignant complications in a cohort of celiac disease (CD) patients under gluten-free diet (GFD). Five hundred and forty-nine CD patients were retrospectively assessed. Two hundred and fifty-one (45.7%) showed classical, 262 (47.7%) subclinical, and 36 (6.6%) silent form of CD at the time of the diagnosis. The mean time under GFD was 7.13 years (range 1–15 years). Out of 549 patients, 381 (69.4%) were fully compliant, 112/549 (20.4%) reported less than one dietary transgression/month, and 56/549 (10.2%) reported at least one dietary transgression/month. Out of 549 patients, 18 (3.3%) patients developed complications under GFD (seven malignant and 11 nonmalignant complications). Fourteen patients were previously affected by classical CD (5.6% of the overall patients with classical CD), and four were affected by subclinical CD (1.5% of the overall patients with subclinical CD). None of the patients affected by silent CD developed complications. There was no statistical difference between the mean age of the two groups developing complications (P = n.s.). Complications appeared after a mean time under GFD of 6.5 years in classical CD, and after a mean time of 3.5 years in subclinical CD (P = n.s.). Finally, 6/14 (42.8%) patients with classical CD were not fully compliant to GFD, while 2/4 (50%) of subclinical CD patients were not fully compliant to GFD (P = n.s.). Less than 5% of CD patients may develop complications under GFD. Complications seem to affect more classical CD than subclinical CD, and seem to be irrespective of optimal GFD adherence.

Keywords

Celiac diseaseComplicationsGluten-free diet

Introduction

Celiac disease is a chronic inflammatory disease of the gut occurring in genetically susceptible individuals after ingestion of gluten. It is characterized by a flattened mucosa, villous atrophy and crypt hyperplasia in the small intestine, and by the classic malabsorption syndrome (diarrhea, steatorrhea, weight loss) or by apparently minor symptoms such as iron-deficiency anemia, osteopenic bone disease, amenorrhea, and infertility [1]. The elimination of gluten from the diet generally leads to a return to normality of the morphological changes [2].

Most gastroenterologists at present recommend a strict gluten-free diet (GFD) in all patients diagnosed with celiac disease. However, this should not be undertaken without proper consideration. GFD may have psychological effects on celiac patients [3], such as poor compliance to GFD itself. A plenty of literature data on malignancy in celiac patients are available [4], as well as literature data on nonmalignant complication in celiac disease under GFD are available on the risk to develop autoimmune diseases, sepsis or fractures [58].

The aim of this study was therefore to assess the incidence of malignant and nonmalignant complications in a cohort of celiac disease patients under GFD, and to assess whether the onset of complications is related to nonadherence to GFD.

Methods

Five hundred and forty-nine celiac disease (CD) patients, enrolled from 1993 to 2006, were retrospectively assessed about the onset of complications under GFD.

Both the original [9] and revised [10, 11] criteria for the diagnosis of celiac disease were used in this study.

Endoscopic Procedures and Classification

In all patients IgA and IgG anti-gliadin (AGA), anti-tissue transglutaminase (anti-tTG), and anti-endomysium (EMA) were evaluated.

Esophagogastroduodenoscopy (Fujinon EG300 videogastroscope; Fujinon, Omiya, Japan) was performed in all patients who underwent small bowel biopsy.

Small-bowel biopsies were obtained at the second duodenal portion during esophagogastroduodenoscopy. At least six duodenal endoscopic specimens were obtained from this portion using a disposable biopsy forceps with spike and evaluated by hematoxylin–eosin coloration. Four additional biopsies were obtained from the duodenal bulb in case of endoscopic aspect of micronodular bulb. Celiac disease was defined as a permanent gluten-sensitive enteropathy, primarily expressed by the presence of characteristic small intestinal lesions [12].

Classical CD was defined by the presence of a gluten-sensitive enteropathy with gastrointestinal symptoms (diarrhea, abdominal pain, weight loss, malabsorption syndrome). Subclinical CD was defined by the presence of a gluten-sensitive enteropathy with extraintestinal symptoms (iron-deficiency anemia, alopecia, recurrent abortion, etc.) and without gastrointestinal symptoms. Silent CD was defined by the presence of a gluten-sensitive enteropathy not accompanied by any symptoms, but identified during the course of screening of high-risk groups (first-degree relatives of celiac patients, patients with insulin-dependent diabetes, Down’s syndrome, IgA deficiency, and thyroid disorders) [1].

Histological Classification

Histopathology was expressed according to the Marsh modified classification [13, 14]. For the cutoff for intraepithelial lymphocytes, we used 25/100 according to Hayat et al. [15] and Dickson et al. [16]. Other possible causes of villous atrophy or duodenal damage, such as Giardia lamblia infestation, tropical sprue, collagenous sprue, food protein hypersensitivity (cow’s milk, eggs, fish, rice, chicken) were excluded, as well as any other causes of inflammatory infiltration of duodenum, such as peptic duodenitis, were excluded before making diagnosis of celiac disease [17].

Compliance to Gluten-Free Diet

After celiac disease diagnosis was made, the patients underwent gluten-free diet (GFD) and were reassessed about compliance to GFD during long-term follow-up. Compliance to GFD was assessed according to an arbitrary quantitative scale and based on patient interview at the time of the control visit (every year or at the time of the appearance of the complication):
  • No dietary transgression

  • Less than one dietary transgression/month

  • More than one dietary transgression/month

Statistical Methods

Data analysis was carried out by using Fisher’s exact test with Yate’s correction for small numbers, Student’s t-test for unpaired data, Mann–Whitney two-samples U-test, as appropriate. Values of P < 0.05 were considered as statistically significant differences.

Results

Two hundred and fifty-one patients (45.72%) showed classical, 262 (47.72%) subclinical, and 36 (6.56%) silent form of CD.

The mean time under GFD was 7.13 years (range 1–15 years). Out of 549 patients, 442 (80.5%) underwent endoscopic-histological follow-up of the disease. One hundred and seven patients (19.5%) refused any further intestinal biopsies, and were followed-up by serological tests (Table 1).
Table 1

Histology and serology in CD patients who did or did not develop complications under GFD

 

Patients without complications

Patients developing complications

Histology at the beginning of GFD

    Marsh IIIc

103/531 (19.4%)

6/18 (33.4%)

    Marsh IIIb

142/531 (26.8%)

7/18 (38.9%)

    Marsh IIIa

135/531(25.4%)

4/18 (22.2%)

    Marsh II

116/531 (21.8%)

1/18 (5.5%)

    Marsh I

35/531 (6.6%)

0/18 (0%)

Histology at the end of follow-up (or at the time of the appearance of complication)a

    Marsh IIIc

7/442 (1.6%)

8/18 (44.5%)

    Marsh IIIb

9/442 (2%)

4/18 (22.2%

    Marsh IIIa

38/442 (8.6%)

5/18 (27.8%)

    Marsh II

46/442 (10.4%)

1/18 (5.5%)

    Marsh I

49/442 (11.1%)

0/18 (0%)

    Marsh 0

293/442 (66.3%)

0/18 (0%)

    n.a.

107 patients

Serology at the beginning of GFD

    AGA +ve

376/531 (70.8%)

17/18 (94.2%)

    EMA +ve

451/531 (85%)

18/18 (100%)

    Anti-tTG +ve

462/531 (87%)

18/18 (100%)

Serology at the end of follow-up (or at the time of the appearance of complication)

    AGA +ve

73/531 (13.7%)

n.a.

    EMA +ve

41/531 (7.7%)

5/18 (27.8%)

    Anti-tTG +ve

49/531 (9.9%)

5/18 (27.8%)

    n.a.

3 patients

CD, celiac disease; GFD, gluten-free diet; EMA, anti-endomysium antibodies; anti-tTG, anti-tissue transglutaminase; n.a., not assessed

aFor further explanation about the follow-up of the patients see the text

Of patients undergoing endoscopic-histological follow-up, 256/442 (57.90%) of them normalized histology within 2 years. Fifty-three patients (12%) underwent further intestinal biopsy between 3 and 5 years after CD diagnosis was made: 37 of them (8.4%) showed normal histology of the small bowel, while 16 (3.6%) showed persisting Marsh I lesions at this time. Finally, 133 (13.7%) refused further intestinal biopsies despite persisting histological lesions (Marsh I-IIIa) (Table 1).

Looking at compliance to GFD, 381/549 patients (69.4%) were fully compliant, 112/549 patients (20.40%) reported less than one dietary transgression/month, and 56/549 patients (10.20%) reported at least one dietary transgression/month (Table 2).
Table 2

Clinical form of the disease, and dietary adherence in CD patients, who did or did not develop complications under GFD

 

Overall population

Patients without complications

Patients with complications

Classical CD

251/549 (45.7%)

237/531 (44.6%)

14/18 (77.8%)

Subclinical CD

262/549 (47.7%)

258/531 (48.6%)

4/18 (22.2%)

Silent CD

36/549 (6.6%)

36/531 (6.8%)

0/18 (0%)

Fully compliant to GFD

381/549 (69.4%)

371/531 (69.9%)

10/18 (55.5%)

<1 dietary transgression/month

112/549 (20.4%)

108/531 (20.3%)

6/18 (33.3%)

>1 dietary transgression/month

56/549 (10.2%)

54/531 (10.1%)

2/18 (11.2%)

CD, celiac disease; GFD, gluten-free diet

Looking at the complications recorded during follow-up, 18/549 (3.27%) patients developed complications under GFD (Table 3): 14 were previously affected by classical CD (5.58% of the overall patients affected by classical CD), while four were previously affected by subclinical CD (1.53% of the overall patients affected by subclinical CD). None of the patients suffering from silent disease developed complications.
Table 3

Characteristics of the CD patients developing complication under GFD

Sex

Histology at diagnosis of CD

Clinical form of CD at diagnosis

HLA

Age at diagnosis of complication (years)

Years under GFD

EMA at complication

Anti-tTG at complication

Compliance to GFD

Complication

F

IIIc

Classical

DQ2

35

5

+

+

No (>1 transgression/month)

Duodenal adenocarcinoma

M

IIIc

Classical

DQ2

45

5

+

+

No (>1 transgression/month)

Jejunal adenocarcinoma

F

IIIb

Classical

DQ2

40

6

Yes

Siogrën’s disease

F

IIIc

Classical

DQ2

35

3

Yes

Jejunal EATL

F

IIIa

Classical

DQ8

50

15

n.a.

n.a.

Yes

Myocardial infarction

F

IIIb

Classical

n.a.

50

15

n.a.

n.a.

Yes

Myocardial infarction

F

IIIc

Classical

n.a.

48

10

Yes

Sclerosing cholangitis

F

IIIa

Classical

DQ2

20

3

+

+

No (>1 transgression/month)

Herpetiform dermatitis

F

II

Classical

DQ2

24

3

+

+

No (<1 transgression/month)

Herpetiform dermatitis

M

IIIc

Classical

DQ2

62

1

+

+

Yes

Jejunal EATL

F

IIIa

Classical

n.a.

55

2

Yes

Colonic adenocarcinoma

M

IIIb

Classical

n.a.

62

3

No (<1 transgression/month)

Gastric MALT-lymphoma

F

IIIb

Classical

DQ2

18

15

No (>1 transgression/month)

Ulcerative jejunitis

M

IIIc

Classical

n.a.

48

5

Yes

Severe NASH

F

IIIc

Subclinical

DQ8

23

4

Yes

Recurrent abortion

F

IIIc

Subclinical

DQ2

26

4

Yes

Recurrent abortion

F

IIIa

Subclinical

n.a.

53

2

n.a.

n.a.

No (>1 transgression/month)

Jejunal adenocarcinoma

M

IIIa

Subclinical

n.a.

35

4

No (>1 transgression/month)

Autoimmune thrombocytopenia

CD celiac disease, GFD gluten-free diet, EMA anti-endomysium antibodies, anti-tTG anti-tissue transglutaminase, EATL enteropathy-associated T-cell lymphoma, n.a. not assessed

The most recorded complications were malignancies (7/18 patients, 38.89%). Surprisingly, the most common malignancy was not enteropathy-associated T-cell lymphoma (EATL), but small bowel adenocarcinoma (three cases, 16.66%), followed by EATL (two cases, 11.12%). Other cases of gastrointestinal neoplasia, as colonic carcinoma or gastric-MALT lymphoma, were also recorded (Table 1).

As far as the nonneoplastic complications under GFD are concerned, interestingly some patients developed myocardial infarction (two cases, 11.12%). Other inflammatory complications (severe nonalcoholic steatohepatitis), autoimmune diseases (sclerosing cholangitis, Siogrën’s disease, autoimmune thrombocytopenia) or an unexplained disease (two cases of recurrent abortion) were also recorded.

Finally, one case of ulcerative jejuno–ileitis and two cases of herpetiformis dermatitis were recorded.

Looking at the clinical form of CD affecting these patients, 14 patients were previously affected by classical CD (3 men and 11 women, mean/median age 42.28/46.50 years, standard deviation 5.08 years; 5.58% of the overall patients with classical CD), and four were affected by subclinical CD (two men and two women, mean/median age 34.25/30.5 years, standard deviation 13.5 years; 1.53% of the overall patients with subclinical CD). Surprisingly, none of the patients affected by silent CD developed complications. There was no statistical difference between the mean age of the two groups developing complications (P = n.s.).

Considering the time of appearance of the complication after GFD has started, complication appeared after a mean/median time under GFD of 6.5/5 years (standard deviation 5.08 years) in classical CD, and after a mean/median time of 3.5/4 years (standard deviation 1 year) in subclinical CD. No statistical difference was found between the two groups (P = n.s.).

Finally, looking at compliance to GFD in patients developing complications, we found that only 6/14 (42.85%) patients with previous classical CD were not fully compliant to GFD, while 2/4 (50%) of subclinical CD patients were not fully compliant to GFD (P = n.s.).

Discussion

The majority of studies in this field found that CD people have a higher mortality than general population (standardized mortality ratio ranging from 1.26 to 2.0) [4, 18, 19]. Moreover, childhood CD seems to show higher risk of malignancy than adulthood, but this result seems to be related to external causes and not to the disease [20]. Although mortality in untreated celiac disease is increased, most studies show that this normalizes within 5 years of treatment [21].

The most important cause of mortality in CD is represented by malignancies. EATL is the most common malignancy and the major cause of death [4, 18, 19]. Other gastrointestinal malignancies are recorded in the literature (small bowel, gastric, esophageal, and colonic carcinoma, non-Hodgkin’s lymphomas, etc.), but a limited number of cases were described in every report. This study confirms that the most recorded complications in CD population were malignancies (7/18 patients, 38.9%). Surprisingly, the most common malignancy in our experience was not EATL, but small bowel adenocarcinoma (three cases, 16.66%), followed by EATL (two cases, 11.1%). It is not easy to explain this finding. Small bowel adenocarcinoma was never described as the most frequent malignancy in CD; it is generally described in small percentages, and sometimes represents the first presentation of CD [22]. The most considered hypothesis about the development of small bowel adenocarcinoma in CD is the long-term history of CD. Both Swedish [23] and American experiences [24] support this hypothesis. However, our cases show a CD history ranging from 2 to 5 years; although we cannot exclude a longer history of histological lesions before making diagnosis of CD, the real cause of the higher incidence of this neoplasia in our population remains unknown.

Other cases of gastrointestinal neoplasia, as colonic carcinoma or gastric-MALT lymphoma, were recorded in our population. In particular, the case of gastric MALT-lymphoma is interesting, because of the H. pylori-negative status of this patient. This neoplasia is quite rare in CD [19], and only three cases have been described in the literature [2527]. Acquired mucosa-associated lymphoid tissue (MALT) is the background for the development of that specific subtype of gastric lymphoma, and CD is one of the extragastric diseases in which acquired gastric MALT can be found [28]. Lymphocytes gastritis, a typical gastric T-cell infiltration often founding in CD [29], may be the key point, and may activate the B-lymphocytes. The long-term chronic stimulation of T- and B-cells may lead to acquired MALT and then to gastric MALT-lymphoma.

Looking at other complications occurring in CD under GFD, several cases of autoimmune disease (as Sjögren’s disease or autoimmune thrombocytopenia) were recorded in our population. These diseases are quite rare in CD. Thrombocytopenia has rarely been reported in CD, and may be autoimmune in nature [30]. Siögren’s disease has been reported in CD, and its appearance in CD is probably related to environmental factors as triggers or precipitators [31]. Also a case of severe non-alcoholic steato-hepatitis (NASH) was observed. This patient showed increased hepatic enzymes at the time of diagnosis, as frequently observed in CD [32]. Despite a strict compliance to GFD, and despite a body mass index <25 kg/m2 (normal value), he developed a severe NASH (confirmed by histology) about 5 years after GFD has started. CD and NASH may share similar pathogenetic mechanism resulting from increased tumoral necrosis factor-α and increased permeability [33], leading to translocation of gut bacteria, Kupffer cell stimulation, and reactive oxygen species production, resulting in liver inflammation [33]. This pathogenetic mechanism may be considered in patients at the first diagnosis of CD with associated NASH. In fact GFD generally obtains normalization of liver blood test abnormalities, as well as reversal of histological abnormalities, within 6 months after GFD has started [34]. We do not know why this patient developed that complication. Our patient developed NASH after 5 years of strict GFD, while symptoms had disappeared, and he was well. It is therefore hypothesized that intestinal permeability related to CD should have disappeared. Another hypothesis is that this patient could suffer from small bowel intestinal overgrowth, a condition often diagnosed in CD [35]. Intestinal permeability related to this condition may permit persistence of events related to increased intestinal permeability and resulting in liver inflammation, but this intriguing hypothesis is supported only by rat model at present [36]. Finally, the last hypothesis is that NASH development may be related to causes other than CD.

Regarding patients who developed recurrent abortion, it is not easy to explain this outcome because mechanisms for altered pregnancy outcome in CD women are only speculative at present. Literature found that gross indices of severity of the disease do not correlate to the outcome of pregnancy [37], suggesting that other selective nutrient deficiency, vitamins or hormones may play a significant role in pregnancy and fetal development [38, 39]. Both our patients were not fully compliant to GFD, which may explain the persistent risk of pregnancy resulting in recurrent abortions.

An interesting complication recorded in this study was the appearance of myocardial infarction in two females under strict GFD. It is unclear why these patients developed this complication. The relationship between CD and vascular disease may be complicated since CD is linked to protective factors for vascular disease, such as lower serum cholesterol levels and indications for lower blood pressure, but also with risks for vascular disease such as lower folate levels and higher homocysteine levels [40]. Homocysteine levels in treated CD patients are only marginally increased [41], but it remains a biologically plausible possibility that low-grade inflammation may result in a modest increase in the risk of vascular disease in those patients [42].

Surprisingly, no neuropsychiatric disturbance was recorded in our population. The prevalence of these types of complications seems to be increasing in celiac population [43]. The absence of these complications could therefore be related to a different mechanisms than simple gluten toxicity, such as different genetic susceptibility.

Two interesting findings may be explain our results. The compliance to GFD, a key factor in reducing the neoplastic risks in CD [44], was not optimal in our population. In fact less than 70% of our patients were fully compliant to GFD, while about 30% of them reported different degrees of noncompliance to the diet. These data are similar to those described in other larger studies, in which good compliance to GFD ranged from 59% to 80% [18, 19]. Poor compliance to GFD may therefore be a key factor in explaining our data. But surprisingly neither dietary compliance nor the time under GFD seemed to be adequate factors to explain our data. In fact we found that about 42% of patients with previous classical CD and 50% of patients with previous subclinical CD were not fully compliant to GFD, without any statistical differences. As far as time under GFD at time of appearance of complication is concerned, complications appeared after a mean/median time under GFD of 6.5/5 years (SD 5.08 years) in classical CD, and after a mean/median time of 3.5/4 years (SD ± 1 year) in subclinical CD, without any statistical difference between the two groups (P = n.s.). Dietary compliance therefore does not seem to be a strong risk factor in developing complication under GFD.

On the contrary, the clinical form of the disease at the time of diagnosis of CD seems to be more important in determining the development of complications. In fact 5.6% of the overall patients affected by classical CD developed complications under GFD, while only 1.5% of the overall patients affected by subclinical CD developed complications under GFD. Our data are too limited to perform an adequate statistical analysis, but they seem to be indicative for a risk of complications related to the classical presentation of the disease. Literature defines classical CD as at higher risk of severe endoscopic and histological damage [45], as well as more frequently affected by malabsorption than subclinical or silent form of CD [46]. Duration of gluten exposure may be a key point to explain the different histological damage with consequent different clinical appearance of the disease. When CD is diagnosed early it is probable that gluten had not yet caused serious damage, with consequent slight/mild histological damage associated with subclinical or silent disease at the time of the diagnosis. On the contrary, in older patients gluten would cause prolonged damage which would show itself as more severe histological damage with consequent classical, active disease [47]. According to these data, it is therefore hypothesized that those patients suffering from classical CD may be at higher risk of complications over patients suffering from subclinical/silent CD, and our data seem to support this hypothesis.

Conclusion

We found that less than 5% of CD patients may develop complications under GFD. Complications seem to be irrespective of optimal GFD adherence, and seem to affect more patients with previous diagnosis of classical than subclinical CD. If these data are confirmed by others, patients affected by classical CD should be strictly followed up for higher risk of complications even if strictly compliant to GFD.

Copyright information

© Springer Science+Business Media, LLC 2008