Celiac Disease: Could You Be Missing This Diagnosis?

The American Gastroenterological Association Technical Review on Celiac Sprue defines CD as "a condition in which there is an abnormal proximal small intestinal mucosa that improves morphologically on treatment with a gluten-free diet and relapses when gluten is reintroduced."¹ Serologic testing is extremely helpful in making the diagnosis, but a histologic diagnosis based on duodenal or jejunal pathology and the response to a gluten-free diet (GFD) are essential. Rechallenge with gluten is usually unnecessary.

In this review, I offer practical suggestions that address the principal components of diagnosis and management.

OVERVIEW

In the 1940s, a pediatrician in Utrecht, Netherlands, noticed the dramatic improvement in CD symptoms among Dutch children during World War II as a result of the bread shortage, which reversed when grains were reintroduced. Researchers subsequently found that wheat, barley, rye and, to a lesser degree, oats triggered a malabsorption syndrome that resolved with the elimination of these products. Further work isolated the toxic agent in gluten, the alcohol(-soluble fraction of wheat protein. Since the 1990s, our understanding of the genetics and pathophysiology of CD and our diagnostic accuracy have advanced dramatically.

CD is neither a food allergy nor a food intolerance—a concept that may confuse some patients. Although chronic skin diseases, such as dermatitis herpetiformis, are linked with CD, there is rarely an associated or immediate urticarial reaction in response to gluten ingestion, such as is seen in allergy. Unlike a food intolerance, such as lactose or fructose intolerance, which usually produces GI symptoms within hours of exposure, the mucosal changes of CD occur over months to years and are not temporally related to day-to-day dietary exposure.

EPIDEMIOLOGY

The prevalence of CD has historically been underestimated. During the past decade, known prevalence has risen to about 1 in 200 Americans.^2-4 Persons of all ages are affected. The female to male ratio is about 2:1. CD is less common in non-white persons and is most frequently seen in Europe and in countries to which Europeans have emigrated (such as the United States and Australia). First-degree relatives of affected persons have a roughly 10-fold risk of CD, and there is a more than 70% concordance in identical twins. This evidence supports the notion that although there is a dietary trigger, CD has a clear genetic basis.

CLINICAL PRESENTATIONS

The diagnosis of CD is frequently missed. The protean manifestations of the disease, many of which are extraintestinal, may bring patients to a primary care clinician, neurologist, gynecologist, hematologist, or rheumatologist rather than to a gastroenterologist. Only a small minority of patients present with the textbook symptoms of malabsorptive diarrhea with steatorrhea, weight loss, and nutritional deficiencies (usually folate and iron). Many persons have a subclinical enteropathy. Others have GI complaints without constitutional symptoms, which may suggest irritable bowel syndrome.⁵ Others may have persistent travelers' diarrhea.⁶

The numerous extraintestinal manifestations and conditions associated with CD are listed in Table 1.^7-9 Some extraintestinal manifestations, such as dermatitis herpetiformis, have a causative relationship with CD; 85% of patients with this skin condition have CD on duodenal biopsy, almost all are serologically positive for CD, and most respond to a GFD. Other disorders, such as autoimmune thyroid disease and type 1 diabetes, may be genetically linked, but usually do not improve with a GFD.^10,11

CD is associated with a host of neurologic sequelae—including neuropathy, ataxia, and epilepsy—the neuropathophysiology of which we are just beginning to understand.¹² In one study, the incidence of malignancy, the most feared complication of CD, was 1.3 compared with the incidence in the general population, with a commensurate rise in mortality from malignancy.¹³ (However, with extended follow-up, the 1.3 figure did trend toward unity.) Cancer was diagnosed in 249 of 12,000 persons with CD or dermatitis herpetiformis who were followed from 1964 to 1994. The incidence of malignant lymphoma showed the most dramatic increase and accounted for 18% of cancers. The incidence of esophageal squamous cell carcinoma, duodenal and colonic adenomas, and hepatocellular carcinoma also increased modestly.

Continued on next page

DIAGNOSIS

History and physical examination. If CD is suspected, a thorough history taking and physical examination are in order. Carefully question the patient about GI symptoms, such as bloating, abdominal pain, chronic diarrhea and/or irritable bowel symptoms. An extensive review of systems is warranted and should focus on constitutional signs (such as weight loss and fatigue) and extraintestinal manifestations (such as skin lesions and oral aphthae). Highlights of the physical examination include inspection for de-enamelization of the teeth, muscle atrophy, and kyphosis. Although most patients with active CD tend to be thin, overweight patients may have latent or active CD; this diagnosis should be considered in the appropriate clinical setting.

Routine laboratory testing. Routine laboratory tests include a complete blood cell count; mean cell volume; red blood cell distribution width; and iron, vitamin B₁₂, and folate levels. Because CD is largely a disease of the proximal small bowel, patients typically have low folate and iron levels but normal B₁₂ levels. (This is in contrast to tropical sprue, a disease with a similar phenotype, which involves the entire small bowel, including the ileum, and often produces both B₁₂ and folate deficiency.) Erythrocyte sedimentation rates may be normal or mildly elevated. A finding of hypoalbuminemia may indicate a protein-losing enteropathy. In patients with chronic diarrhea, stool should be examined for fecal fat (at least a spot test, if not a proper 72-hour collection), ova, and parasites and tested for Giardia antigen and Clostridium difficile.

Serologic testing. The availability of accurate serologic tests has greatly facilitated the diagnosis of CD. The principal features of these tests are reviewed in Table 2.

In most cases, it suffices to obtain tests for IgG and IgA antigliadin antibodies (AGA), enzyme tissue transglutaminase antibodies (tTGA), and total IgA (to rule out false negatives from IgA deficiency, which is common in patients with CD).¹ The low specificity of the AGA tests has led some experts (including the authors of the NIH consensus statement) to discourage their use. However, there is a small minority of patients with CD who are AGA-positive and tTGA-negative.

The IgA antiendomysial (EMA) test is slightly more specific than the tTGA tests and is used to help confirm the diagnosis. In a patient with a low clinical likelihood of CD, the diagnosis may be reasonably excluded with negative results on serologic testing (tTGA or EMA with or without an AGA test and normal total IgA level). In a patient with a high clinical likelihood of CD, negative results on serologic testing do not rule out the diagnosis, and duodenal biopsies must be obtained. Duodenal biopsies are required in all cases to make a definitive diagnosis.

One could argue that duodenal biopsy is not needed when the IgG AGA test is the only positive serologic study because its specificity is so poor. Negative results on a sensitive but nonspecific test like IgG AGA in a patient who is unlikely to have CD (such as a non-white person with abnormal liver chemistries) are useful to rule out the diagnosis. However, positive results in such a patient have an extremely low positive predictive value and are often falsely positive. A specific test such as the EMA—when results are positive in a patient who is very likely to have the disease (such as a patient of European ancestry with a family history of CD and malabsorptive diarrhea)—has a nearly 100% positive predictive value. In the same high-risk patient, negative results on EMA testing may simply be wrong.

Duodenal biopsy. Typical endoscopic findings in CD include mucosal atrophy, fissuring, and scalloping (where fissures cross over folds) (Figure 1). Histopathologic examination of a biopsy specimen typically reveals villous atrophy and crypt hyperplasia (Figure 2), but in the early stages may demonstrate only nonspecific lymphocytic infiltration with normal villous architecture.

The argument is often madeto treat presumptively for CD based solely on positive results of serologic tests, without benefit of a duodenal biopsy. I strongly recommend against this approach for several reasons:

• Histologic assessment remains the gold standard for diagnosis.
• Clinical criteria are nonspecific.
• Upper endoscopy is a safe and simple procedure.
• False-positive serologic test results can lead to an impaired quality of life and the unnecessary implementation of a life-altering diet.
• Symptomatic improvement with a GFD is not specific and is susceptible to placebo effect.
• It is much harder to rule the diagnosis in or out after a GFD because of potential normalization of serologic studies and biopsies.

A caveat in endoscopic diagnosis is that the grossly normal appearance of duodenal mucosa does not obviate the need for biopsy. Gross mucosal signs, such as mosaicism or scalloping, are only 50% sensitive compared with the gold standard of histologic assessment.¹⁴ However, histologic evaluation is also subject to error in cases of inadequate sampling (at least 6 biopsies from nonbulbar duodenum should be taken), a non-reported GFD, and inadequate pathologist experience.

Genetic testing. Diagnosis is more difficult in a patient who has self-initiated a GFD, because results of serologic tests and duodenal histologic assessments may revert to negative. In these patients, a gluten rechallenge may be instituted, followed by repeated serologic testing and biopsies. Because many patients are reluctant to undergo these steps, a genetic screen for DQ2/DQ8 may be sent. This test is considered necessary but not sufficient for the diagnosis. HLA-DQ2 and/or HLA-DQ8 are found in at least 95% of patients.¹⁵ Therefore, unless the likelihood of disease is extraordinarily high, a negative result largely rules out the diagnosis. A positive result, however, sheds no light, because the alleles are carried in approximately 30% of the population.

MANAGEMENT

Treatment consists of strict adherence to a GFD. Products that contain wheat, barley, and rye must be avoided. Oats are generally forbidden as well because of contamination by other grains in production. A GFD is extremely effective, albeit difficult because of the centrality of grains to the Western diet and the widespread presence of gluten in processed foods. However, patients usually experience such a dramatic improvement in GI symptoms and in their general sense of well-being that most do not complain about compliance. Dietitians and CD societies' Web sites can help direct patients to appropriate products, restaurants, and support groups.

If a patient does not respond to a GFD within 3 months, the most likely cause is continued gluten exposure, of which the patient may or may not be aware. An IgA AGA level that does not fall is clear evidence of such exposure.Other diagnoses to consider include lymphocytic colitis, lactose intolerance, pancreatic insufficiency, unrelated irritable bowel syndrome, drug-induced diarrhea, Crohn disease, ulcerative colitis, refractory or collagenous sprue, and ulcerative jejunitis (a rare entity with poor prognosis). The occasional patient with refractory CD requires corticosteroids, but the focus generally should be on scrutinizing the diet. There may also be a role for probiotics, which have protean effects on mucosal immunity as well as the capacity to hydrolyze gliadin polypeptides.¹⁶ However, data are preliminary.

Other management considerations include repletion of iron, folate, and calcium (vitamin B₁₂ is usually unaffected). Screen patients for associated disorders such as thyroid disease and diabetes mellitus. Bone densitometry and age-appropriate cancer screening are reasonable. Routine small-bowel imaging by small-bowel series or capsule enteroscopy is controversial. Pneumococcal vaccination is recommended because of the prevalence of hyposplenism among patients with CD.

SCREENING

Most experts do not recommend screening of asymptomatic persons, except perhaps for first-degree relatives of persons with CD. However, this must be sharply distinguished from screening in patients with GI symptoms or possibly related extraintestinal signs or symptoms, especially anemia, fatigue, premature osteoporosis, and the associated conditions listed in Table 1. In these settings, pursuing the diagnosis is strongly encouraged to reduce morbidity. *