Association of Adult Celiac Disease With Surgical Abdominal Pain

Abstract

Background:

Acute abdominal pain is the most common indication for surgical admission. Nonspecific abdominal pain (NSAP) may account for up to 40% of cases. There has been no published prospective study in which adult patients presenting with acute abdominal pain are investigated for celiac disease.

Aims:

We aimed to assess the association of celiac disease with surgical abdominal pain.

Patients and Methods:

A case-control study was undertaken involving 300 consecutive new unselected patients presenting with acute abdominal pain (in a university hospital) and healthy controls (age and sex matched) without abdominal pain (n = 300). Initial investigations for celiac disease were immunoglobulins, IgA/IgG anti-gliadin (AGA), and endomysial antibodies (EMA). Any patient with a positive IgA AGA, EMA, or only IgG AGA in the presence of IgA deficiency was offered a small bowel biopsy to confirm the diagnosis.

Results:

There were 33 patients with abdominal pain who had positive antibodies, of whom 9 had histologically confirmed celiac disease (6 EMA positive; 3 EMA negative). One antibody positive patient (EMA in isolation) declined duodenal biopsy and the remaining 23 had normal duodenal mucosa. Within the control group, there were 2 cases of celiac disease. Compared with matched controls the association of acute abdominal pain with celiac disease gave an odds ratio 4.6. (P = 0.068, 95% confidence interval, 1.11–19.05). When only considering NSAP the prevalence of celiac disease was highly significant at 10.5% (9 of 86, P = 0.006). Patients’ symptoms improved on a gluten-free diet at 12- to 18-month follow-up.

Conclusion:

Celiac disease was diagnosed in 3% of patients who presented with unselected acute abdominal pain to secondary care. Targeting patients who have NSAP or celiac associated symptoms/diseases may improve the diagnostic yield.

This is the first study to report a relationship between adult celiac disease and acute abdominal pain. Patients admitted with acute abdominal pain who are subsequently considered to have nonspecific abdominal pain should be investigated for celiac disease.

Acute abdominal pain accounts for up to 50% of emergency surgical admissions.¹ Computer-aided diagnostic questionnaires,² abdominal ultrasound,³ computed tomography (CT) scan of the abdomen,^4,5 early laparoscopy,^6,7 and peritoneal aspiration (with neutrophil counting)⁸ have all been described as potential methods for improving the diagnostic yield in this group of patients. However, despite these advances, a cause for the patients’ symptoms may not always be recognized. As a result, surgeons have internationally described the entity of nonspecific abdominal pain (NSAP).^9–12 This is defined as “pain for which no immediate cause can be found (during the acute admission) and specifically does not require surgical intervention.”⁹ NSAP may be self-limiting and accounts for 13% to 40% of all surgical admissions.^9–12 Follow-up of this cohort of patients has revealed diverse clinical outcomes. De Dombal (who originally classified this entity)⁹ reported that up to 10% of patients with NSAP were subsequently found to have an intra-abdominal malignancy (if over the age of 50 years).¹³ Gynecologic causes of NSAP have also been recognized.^11,12 Conversely, other investigators have suggested a more favorable outcome, with 77% of patients with NSAP being symptom free, at the 5-year follow-up.¹⁰ More recently, an association between NSAP and irritable bowel syndrome (IBS) has been described suggesting that there may be a functional aspect to NSAP.¹⁴ IBS is considered to be a functional gastrointestinal disorder and is defined by symptoms (fulfilling the Rome II criteria) in the absence of known structural pathology.

Abdominal pain may be a presenting feature of celiac disease.^15–18 The prevalence of adult celiac disease in the general population has been reported to be in the magnitude of 1 per 100 to 200. This has been determined by epidemiologic studies screening cohorts of healthy volunteers in the United States, United Kingdom, and other European countries.^19–25

Celiac disease or gluten sensitive enteropathy is defined as a state of heightened immunologic responsiveness to ingested gluten (from wheat, barley, or rye) in genetically susceptible individuals.^24,27,32 Patients with celiac disease may initially be recognized by using noninvasive serologic tests. Reported sensitivities and specificities of these antibodies are variable (IgG anti-gliadin antibody [AGA]: sensitivity, 30%–50%; specificity, 30%–70%; IgA AGA: sensitivity, 50%–70%; specificity, 50%–85%; endomysial antibody [EMA]: sensitivity, 70%–90%; specificity, 80%–100%).^24,27,32 However, the positive predictive value of these combined serologic markers (IgG/IgA AGA and EMA) is in excess of 95%.^24,27,32 In the presence of a positive antibody, the diagnosis of celiac disease should be confirmed by performing a duodenal biopsy. Histologic demonstration of small bowel villous atrophy remains the gold standard for making the diagnosis of celiac disease. International guidelines now suggest that assessment of compliance to a gluten-free diet may be based on patients’ antibody profiles becoming negative after the institution of a gluten-free diet (suggesting no exposure to gluten).^24,27,32

Historically, patients with adult celiac disease typically complained of gastrointestinal symptoms suggestive of malabsorption. This manner of presentation is now described as the classic (typical) form. The increasing recognition of this condition is attributed to novel serological assays and the realization that patients do not always have gastrointestinal symptoms (silent or atypical form) but may present insidiously, for example, with iron deficiency anemia, osteoporosis, cryptogenic hypertransaminasemia, or neurologic symptoms.23,24,26,27

We have previously described an association between celiac disease and IBS.²⁸ In addition, we demonstrated that patients’ symptoms improved on a gluten-free diet.²⁹ When we retrospectively assessed how patients with celiac disease presented in our center, we reported that 16.3% had abdominal pain and many had previously been seen in surgical departments or had a gastroscopy without duodenal biopsy prior to their diagnosis.¹⁸ The mean delay in diagnosis in our cohort was 4.9 years (range, 0.25–16 years), and this is consistent with the international literature.^17,24 Other investigators have reported that patients with undiagnosed celiac disease are more likely to have surgical interventions.³⁰ Despite celiac disease affecting up to 1% of the population, a recent meta-analysis suggested that the ratio of known to undiagnosed celiac disease was 1 to 7.²⁴ This suggests a failure in case finding for celiac disease, which is well described.^24,28,31 For these reasons, we hypothesized that patients presenting with surgical abdominal pain could have unrecognized celiac disease. There have been no previous controlled studies evaluating this approach.

METHODS

Participants

This was a collaborative study between the University Departments of General Surgery and Gastroenterology. This study was undertaken at a single University Hospital in South Yorkshire, United Kingdom (serving a population of approximately 250,000). This hospital provides both secondary and tertiary services. There is a 24-hour general surgical on-call for emergencies. Data were collected from November 2002 to November 2003. During this time frame, all consecutive unselected patients who presented with acute abdominal pain were included in the study. These patients all had consistent baseline investigations. In our center, we perform full blood count, urea, electrolytes, amylase, C reactive protein, and random glucose. Urinalysis and erect chest and abdominal radiographs are performed selectively according to the clinical assessment. Further investigations including contrast enema, abdominal/pelvic ultrasound, or CT are only arranged following senior review.

To investigate for celiac disease, we used immunoglobulins, IgG/IgA AGAs, and EMA.

We recruited 591 volunteers as part of a screening study for the prevalence of celiac disease in the general population.²⁰ These individuals were recruited from local general practices and served as age- and sex-matched controls. These participants did not receive any financial remuneration for taking part in this study. The volunteers were serologically tested as outlined in the acute abdominal pain population. These were asymptomatic individuals (as proven by a validated gastrointestinal questionnaire),²⁰ who were initially approached in GP waiting rooms. Volunteers were eligible if they were not waiting to be seen as patients. Often they were accompanying patients who may have been their spouse, child, relative, or friend. From this cohort, 300 volunteers were matched for sex and age (by year of birth). Using this approach, there were 3 cases where we could not match controls to the identical year of birth. These 3 cases were at the lower and upper limits of the age distribution. In 1 patient (age, 95 years), the nearest control was 10 years apart and for 2 patients (age, 16 years) the nearest control was 2 years older. At the time of matching cases to controls, we are unaware of the antibody results in both groups.

Ethical approval was obtained from the South Sheffield Research and Ethics Committee. The potential significance of a positive test was explained to all participants and consent obtained.

Laboratory Testing

IgG and IgA AGAs were detected by ELISA. Gliadin from wheat extract (Sigma) was bound to the surface of the wells of a 96-well microtitre plate. After washing dilutions of patients’ serum, calibrants and control serum were added to the wells and the plates were incubated at 37°C for 90 minutes. The plates were then washed and rabbit anti-human IgG and IgA serum conjugated with alkaline phosphatase was added to the wells. After further incubation and washing, phosphatase substrate was added and the color development read on a spectrophotometer. Positive antibody responses were defined as those serum samples that demonstrated more than 0.85 optical density units of color development. The optical densities had to fall within predefined limits for acceptance of the assay.

Anti-EMA was detected by indirect immunoflourescence using monkey esophagus substrate and FITC conjugated antihuman IgA (alpha chain specific, monkey absorbed) antibody.

Quality controls are considered in 2 ways: 1) internal quality controls with the inclusion of positive and negative control sera in every analytical batch; and 2) external quality assessment where the laboratory participates in the National EQA program for these analytes and currently has a perfect score.

Small Intestine Biopsy

Any patient or control with a positive IgA AGA, EMA, or only IgG AGA in the presence of IgA deficiency was offered a small bowel biopsy to confirm the diagnosis of celiac disease.

Distal duodenal biopsies were obtained using biopsy forceps through a conventional forward viewing endoscope (Key-Med). Four specimens were taken from the third part of the duodenum. Samples were initially stored in 10% formalin. Thereafter they were embedded in paraffin wax cut on microtomes at 4 μm and stained with hematoxylin and eosin.

Patients were considered to have celiac disease if they had histologic evidence of villous atrophy in the absence of another cause (grade 3 in accordance with the revised Marsh criteria).³²

Statistical Methods

Statistical analyses of data were performed using SPSS (version 10.0 for Windows). The test of significance used was McNemar's Test for the comparison of the paired populations. The odds ratio was measured by the ratio of the discordant pairs. The confidence intervals for the odds ratio were calculated using Miettinen's test-based approach. The relationship between celiac disease and NSAP was assessed using a Yates-corrected χ². Our data were subsequently reanalyzed by a bio-statistician from the Sheffield University Statistics Unit.

RESULTS

Demographics, Investigations, and Diagnoses

Within the acute abdominal pain cohort, the median age was 57 years (range, 16–95 years) and 150 patients were female. These patients were then age and sex matched blindly with controls. The median age of the control cohort was identical at 57 years (range, 18–85 years) and the sex ratio was also 1 to 1.

All 300 patients had an initial biochemical, hematologic, and immunologic profile. Further investigations performed on this group of patients comprised of 62% who had urinalysis (186 patients) and 56% (168 of 300) who had either an erect chest or abdominal x-ray. Thereafter, 8% (24 of 300) had a contrast enema performed, 54% (162 of 300) had an abdominal/pelvic ultrasound, and 16% (48 of 300) a CT of the abdomen and pelvis. Forty-four patients required surgical intervention (14.7%).

Medical causes of abdominal pain accounted for 9.7% of the group (29 of 300). These cases comprised of pneumonia/chest infection (n = 14), suspected or proven pulmonary embolus (n = 6), presenting metastatic cancer (n = 4), myocardial infarction or cardiac ischemia (n = 3), and 2 cases of decompensated liver disease with abdominal pain from ascites (n = 2). Gastroenteritis was only diagnosed if stool culture positive (n = 3). Functional abdominal pain was only diagnosed in cases (n = 3) where the patient had previously had investigations (prior to this admission), which were normal. In addition, these patients’ symptoms fulfilled internationally established criteria for a functional gastrointestinal diagnosis (all 3 cases had previously normal lower gastrointestinal tract investigations).²⁸

The final diagnosis was considered to be NSAP in 28.7% of patients (Table 1). This diagnosis is specified by the investigating surgeon at the time of discharge and represents the group of patients who did not require surgical intervention and for which no immediate cause for their acute abdominal pain could be found (during the acute presentation or admission).⁹ At the time of specifying this diagnosis, the investigating team is unaware of the results of the celiac antibody profile.

TABLE 1. Final Diagnoses of the Patients Presenting With Acute Abdominal Pain

Diagnosing Celiac Disease

There were 33 patients with abdominal pain who had either a positive IgA gliadin antibody, EMA, or a combination; of these, 9 had celiac disease (Table 2; 6 EMA positive, 3 EMA negative). There was 1 antibody-positive patient (EMA in isolation) who declined duodenal biopsy and 23 who had normal duodenal mucosa.

TABLE 2. Duodenal Biopsy Findings in Patients and Controls With Positive Celiac Antibody Results

There were 2 cases of celiac disease from the controls. One was EMA-positive only and the other IgA gliadin in isolation, a third case was only IgA anti-gliadin positive but had normal duodenal mucosa (Table 2).

Overall, celiac disease was diagnosed in 3% of 300 patients who presented with acute abdominal pain. All the cases of celiac disease had presented with NSAP. This prevalence of celiac disease was 10.5% when considering NSAP exclusively (9 of 86). Within the combined cohort of 600 patients, there were 5 cases of IgA deficiency (IgA < 0.8 g/L): 3 among the abdominal pain group and 2 within the controls. None of these patients had any positive antibody titers, and specifically not IgG anti-gliadin.

After matching cases to controls, we then ascertained the antibody status of both groups and whether they were subsequently found to have celiac disease. Using this approach, there were no concordant pairs. Statistical assessment of the chance of having celiac disease when presenting with acute abdominal pain gave an odds ratio 4.6. (P = 0.068; 95% confidence interval, 1.11–19.05).

The prevalence of celiac disease in patients with NSAP was highly significant with a P = 0.006 (Yates-corrected χ²).

Follow-up of the 9 cases of celiac disease occurred regularly (approximately 3-monthly) for a period of 12 to 18 months. Patients were asked if their symptoms had resolved while adhering to a gluten-free diet. All patients suggested a subjective improvement with resolution of their symptoms. Other than case 2, all of the other patients now had negative antibody profiles (used as markers of compliance to a gluten-freediet).^24,27,32 Case 3 also required antispasmodics (Table 3).

TABLE 3. Celiac Disease Patients: Characteristics and Immunology Results

DISCUSSION

This is the first study to assess the association of celiac disease with acute abdominal pain. We have demonstrated that unrecognized celiac disease was the underlying diagnosis in 3% of this group. Although the odds ratio was 4.6, we have not shown a significant relationship between celiac disease and acute abdominal pain when compared with controls. However, we are the first investigators to describe the relationship between NSAP and celiac disease (P = 0.006). The recognition of previously undiagnosed celiac disease in this group of patients has clinical implications for the investigation of patients with acute abdominal pain. This may be a cohort of patients who should be investigated for celiac disease. All of the cases had symptoms (or other diseases) that could be regarded as associated to celiac disease (Table 3). The presence of such associated conditions may help clinicians identify high-risk patients.^22,27,31 This case-finding approach has been applied with good yield in other symptomatic groups, both in primary and secondary care.^22,28,31

Our understanding of celiac disease is evolving. We now recognize typical, atypical, latent, and potential forms.^22–27 Clearly, at present it is a considerably underdiagnosed condition.22,24,27,28,31 The implications to the individual of a missed or delayed diagnosis are the potential complications of osteoporosis, nutritional deficiencies, infertility, and an increased risk of malignancy.^23,24

The mechanism of abdominal pain in celiac disease is not clear. Historically, reports have described celiac disease presenting with pseudo-obstruction¹⁵ (secondary to hypokalemia) or with small bowel obstruction secondary to lymphoma.¹⁶ An association between celiac disease and idiopathic recurrent pancreatitis has also been suggested.³³ However, none of our patients with celiac disease had any indication of these clinical features, and all had a normal serum amylase. Small bowel intussusception was previously considered to be an asymptomatic feature of celiac disease,³⁴ although more recently a few symptomatic cases have been reported^35,36 (we did not perform any small bowel studies). Alternative hypotheses could be related to reversible autonomic dysfunction of the gastrointestinal tract or IBS.^37,28 Our observation is supported by the improvement of these individuals’ symptoms on a gluten-free diet. In addition, there have been similar reports of celiac disease presenting with recurrent abdominal pain in the pediatric population.^38,39

There are limitations to this study. Within the cohort of patients with acute abdominal pain, 1 individual who was EMA positive declined duodenal biopsy. This patient had presented with urologic obstruction secondary to disseminated prostatic malignancy. We would consider that, even if this patient had agreed to duodenal biopsy, his presentation was not related to celiac disease.

We diagnosed 2 cases of celiac disease from 300 healthy volunteers (0.67%). It is possible that people who think they may have celiac disease are more likely to participate in our volunteers study. However, the prevalence demonstrated in our controls is concordant with 2 other prevalence studies of celiac disease within the United Kingdom, conducted by independent investigators.^19,21

A further limitation could be the definitions of NSAP or acute abdominal pain. The cases of celiac disease that were diagnosed all gave histories of acute on chronic abdominal pain, characterized by preceding symptoms (Table 3). However, our entry criteria for the study used the same definitions as previous surgical investigators assessing acute abdominal pain.^9–12 In total, 38% (114 of 300) of our cohort gave a history of intermittent abdominal pain for months to years prior to this acute presentation or admission. This preceding clinical history for patients with acute abdominal pain is also consistent with previous reports.^9–12

This observation does not necessarily need to be viewed as a limitation to the study. Alternatively, given that all the cases of celiac disease had both NSAP and intermittent abdominal pain, specific inquiry during the course of history taking and symptoms of presentation may allow the investigating clinician to recognize cases of celiac disease more easily. Taking this approach may actually increase the yield of diagnosis of cases of celiac disease, if the serological tests were performed specifically on these presenting subgroups (38%, 114 of 300 of the cohort had intermittent abdominal pain; 28.7%, 86 of 300 had NSAP).

The advent of EMA, which is considered to be highly specific, has resulted in many practitioners, for economic reasons, using this antibody in preference to the other available celiac antibodies. In our study, there were 3 cases of celiac disease (that presented with acute abdominal pain) who were EMA negative (33.3%). These individuals would have been misdiagnosed as NSAP without the use of IgA gliadin antibody in combination with EMA (Table 3, case numbers 7–9). Our observations are not unique; a recent study of a cohort of 69 patients with untreated celiac disease reported the overall EMA sensitivity to be 60%. This Dutch group described a correlation between EMA positivity and the severity of the histologic lesion. Patients with total villous atrophy all had a positive EMA antibody; however, in those with partial or subtotal villous atrophy, the sensitivity of EMA was 31% and 70%, respectively.⁴⁰

We did not observe any cases of celiac disease in the presence of IgG gliadin (in isolation) with IgA deficiency. This study has not provided data to support the requirement of immunoglobulins or IgG gliadin as part of the immunologic assessment for celiac disease.

It is possible that some of our findings were coincidental and may not explain these patients’ symptoms; however, the improvement in symptoms, which occurred following a gluten-free diet, would refute this suggestion. This assertion is supported by the negative antibody profiles after commencement on a gluten-free diet, subjective improvement/resolution of symptoms, and a period of follow-up ranging from 12 to 18 months.

CONCLUSION

Celiac disease was diagnosed in 3% of patients who presented with unselected acute abdominal pain to secondary care. Targeting patients who have NSAP or celiac associated symptoms/diseases may improve the diagnostic yield.