65

Symposium : Gastroenterology & Hepatology

Diagnosis of Celiac Disease Shinjini Bhatnagar and Nitya Tandon Department of Pediatrics, All India Institute of Medical Sciences, New Delhi.

ABSTRACT Celiac disease is an immune mediated enteropathy initiated by ingestion of gluten, in genetically susceptible individuals. With changing epidemiology, celiac disease initially thought to affect only Europeans, has been increasingly reported from other parts of the world including India. However, its true prevalence in India is still not known, as the diagnosis is being missed. The gold standards for diagnosis have been characteristic small intestinal mucosal changes on gluten and a full clinical remission on its removal from the diet. Presence of serological antibodies, which disappear on gluten free diet further confirms the diagnosis. The understanding of the histopathology of celiac disease has changed over the years. The small bowel mucosal lesion of celiac disease is an evolutionary process with normal mucosal architecture and an increase in intraepithelial lymphocytes at one end of the spectrum and classical flat mucosa at the other. In the Indian subcontinent celiac disease has a heterogeneous histological presentation and the diagnosis may be missed if it is based only on severe mucosal changes or the serology is not considered when moderate or mild mucosal changes are present. The last two decades have shown that antiendomysical (Anti EMA) and anti tissue transglutaminase antibodies (anti-tTGA) have a sensitivity and specificity of more than 95% to diagnose celiac disease. Anti EMA tests being operator dependent are more liable to errors and anti- tTGA may be preferred for large scale screening. However, the different source of tTGA antigen, varied techniques of production and the use of arbitrary units by different commercial kits can influence the diagnostic accuracy of the anti-tTGA assay. There is a strong genetic association of celiac disease with HLA-DQ2 or DQ8. The presence of HLA-DQ2 hetrodimer in more than 97% of a group of North Indian patients with celiac disease indicates that this population has a similar genetic risk for the disease. HLA DQ2 typing can be used for ruling out celiac disease where the diagnosis is equivocal as it has a negative predictive value of greater than 95%. Given the protean clinical manifestation and the heterogeneous histology a standard algorithm for diagnosis of celiac disease is important. [Indian J Pediatr 2006; 73 (8) : 703-709] E-mail : [email protected]

Key words : Celiac disease; Diagnosis; Serological antibodies; Histology

Celiac disease is an immune mediated enteropathy initiated by ingestion of gluten, in genetically susceptible individuals. It is characterized by lifelong intolerance to gluten which is a mixture of gliadin and related prolamines present in cereals such as wheat, barley and rye. It was initially believed to be a disease predominantly affecting the European population but interestingly the highest prevalence of 5.6% reported in the Saharawis of the Arab-Berber origin living in the Sahara desert, is almost 10 fold higher than that reported from most European countries.1 Over the last decade with the availability of more sensitive and specific diagnostic tools the epidemiology has changed with more cases of celiac disease being reported from other parts of the world. The prevalence of the disease ranging from 0.6 to 0.8% in the Middle East (reported among blood donors or school children) is almost similar to that of Europe.2 The initial reports of celiac disease from India appeared in the

Correspondence and Reprint requests : Dr. Shinjini, Bhatnagar, Department of Pediatrics, All India Institute of Medical Sciences, New Delhi-110029, India. Mobile : 9810190634

Indian Journal of Pediatrics, Volume 73—August, 2006

early sixties and eighties from the wheat eating areas of North India 3-9 and subsequently in children who immigrated from India and Pakistan to England.10-11 In the absence of population based data from India, celiac disease is being reported only from specialized gastroenterology clinics or tertiary hospitals. 12-16 A significant increase of 15.5 cases per year from a North Indian tertiary hospital further establishes that celiac disease is being increasingly recognized over the last decade. 17 However, the true prevalence in many geographical settings including India is still not known, possibly because the diagnosis is being delayed or missed in a large proportion of individuals. The correct diagnosis of celiac disease is essential as it requires life long adherence to gluten free diet (GFD). Early diagnosis and treatment has additional benefits because studies now suggest that delayed diagnosis is associated with increased prevalence of other autoimmune conditions,18 mortality,19,20 and increased risk of osteoporosis and malignancies. 21 It is therefore imperative to have standardized protocols using sensitive and specific tests that can confirm the diagnosis of celiac disease and identify individuals at risk. 703

66

Shinjini Bhatnagar and Nitya Tandon blunting of intestinal villi. More likely, it is assumed that such a lesion may be caused by tropical sprue, persistent enteric infection or infestation, post infectious complications, or protein energy malnutrition with associated small bowel bacterial overgrowth. Recent data have shown that among Indian children with chronic diarrhea confirmed to have celiac disease, classical flat mucosa with crypt hyperplasia was present only in one third of children while a similar proportion had crypt hyperplasia with moderate villous blunting. 15 Interestingly among children in whom the mucosal changes were limited to mild blunting of villi (equal crypt depth and villus height), at least a third were confirmed to have celiac disease by the presence of serological antibodies, clinical response to gluten free diet and a positive gluten challenge. It appears that celiac disease in the Indian subcontinent has a heterogeneous histological presentation and the diagnosis may be missed or delayed if it is based only on severe mucosal changes or the serology is not considered when moderate or mild mucosal changes are present. Latent and potential forms of celiac disease may be considered in individuals who have a normal histology but positive serology.24 There is evidence to suggest that inflammatory cytokines interleukin (IL) 1α, IL-2, IL-4, interferon-γ and tumor necrosis factor-a may be present in those with potential celiac disease and may further help to classify normal biopsies in the presence of positive serology.25,26 Earlier reports from India have shown that the basal IEL counts in the normal Indian population are higher (20-104) than the west (6 to 40).27,28 It is important to define cut-offs for IEL counts in our population that can be used as diagnostic markers for early gluten enteropathy. Immu­ nophenotyping of IEL and the presence of γδ IEL in intestinal biopsies need to be evaluated as diagnostic markers.

The initial stringent diagnostic criteria of the European Society of Gastroenterology and Nutrition (ESPGAN 1969), required three biopsies to document a structurally abnormal jejunal mucosa on gluten intake, clinical and histological remission on GFD and deterioration of the mucosa during gluten challenge. The revised ESPGAN criteria is more workable and is presently being widely used. 22 The two essential criteria required are characteristic small intestinal mucosal changes while on a diet containing adequate gluten and a full clinical remission following GFD. The presence of serological antibodies, which disappear while on GFD, further support the diagnosis and should be an essential component of the diagnostic work-up. HISTOLOGY REVISITED The characteristic histological features are partial to total villus atrophy, crypt elongation, increased intraepithelial lymphocytes (IEL) with a lymphocyte mitotic index greater than 0.2%, absence of epithelial cell brush border, infiltration of lymphocytes, mast cells, plasma cells, and eosinophils in the lamina propria, increased crypt mitotic index and pseudo stratification of epithelial cells with loss of nuclear polarity. The understating of the histopatho­ logy of celiac disease has changed over the years. Marsh23 has proposed that the small bowel mucosal lesion of celiac disease is an evolutionary process with normal mucosal architecture and an increase in intraepithelial lymphocytes (IEL) at one end of the spectrum and the classical flat mucosa at the other (Table 1). The spectrum of histological changes of celiac disease is well recognized in the western population. In developing country settings, celiac disease is not routinely considered as a likely diagnosis in children presenting with chronic diarrhea and a mild to moderate (modified Marsh 1-3b) (Table 1) TABLE 1. Histology Score for Celiac Disease Marsh score23

Modified Marsh score81

0. 1. 2. 3.

0. IEL 4/100 EC IEL, normal villi) 1. >40/100 EC IEL, normal villus & crypt ht 2. >40/100 EC IEL, hyperplastic crypts, normal villi 3a. >40/100 EC IE, mild villus flatterning, ↑ crypt ht 3b. >40/100 EC IEL, mod villus flattening, ↑ crypt ht 3c. >40/100 EC IEL, total flattening, ↑ crypt ht 4. hypoplasia

0 normal infiltrative (↑IEL, normal villi & crypts) hyperplastic crypts (↑ IEL, normal villi) subtotal-total villus atrophy (villus destruction)

TABLE 2. Range of Sensitivities and Specificities from Studies in India and the West9,29-51 Anti-gliadin

IgG IgA AEA tTG

Sensitivity

Specificity

PPV

NPV

India

West

India

West

India

West

India

West

75 80 - 94 90 76

89-100 70-100 57-100 70-100

78 84-92 95 94

38 - 92 87 - 97 91.7 - 100 84 - 100

75 92 96 97

20 - 95 28 - 100 80 - 100 67 - 100

78 94 87 55

73.90 82-95 93-100 87-100

Wide range of values are because of different gold standards, heterogeneous populations and different lab standards.

704

Indian Journal of Pediatrics, Volume 73—August, 2006

67

Diagnosis of Celiac Disease SEROLOGICAL TESTS Over the last two decades sensitive and specific serological tests have been developed that have helped in demonstrating that the prevalence of the disease is much higher than believed previously. The first immunological assay to be tested were the anti-gliadin antibodies (AGA) which are predominantly of the IgG and IgA class. Solid phase ELISA using crude gliadin as coating antigen has been preferred for detection as there is no evidence to suggest that highly purified wheat preparations improve specificity. Post infection malabsorption, Crohn’s disease and cow’s milk protein intolerance can give false positive tests for AGA. The variability and generally lower accuracy associated with the AGA tests make them unsuitable for diagnostic or screening purposes. There is sufficient evidence to show that IgG-based AGA tests generally are poor in both parameters whereas the IgA-based tests are poorly sensitive but more specific. Over the years these assays have been replaced by more sensitive and specific anti endomysial (EMA), antireticulin (RA) and anti tissue transglutaminase antibodies (tTGA) (Table 2).9, 29-51 Anti-EMA and anti-RA belong primarily to the IgA class and are directed against the intermyofibril substance of the smooth muscle. While the anti-RA react with human and rodent tissue, anti-EMA are species specific reacting only with the gastro-intestinal tract of primates.12 Anti-EMA are identified using indirect immunofluorescence on frozen sections of monkey esophagus. These are available commercially, or home made slides can be prepared using the lower-third of esophagus obtained from Macaque monkeys. Positive serum samples react with the connective tissue which surrounds the smooth muscle bundles in the lamina propria, and give a characteristic lacework pattern around the unstained smooth muscles cells.12 There is a good correlation between human umbilical cord sections and monkey esophagus as substrate for the IgA EMA immune assay and are a cheaper and more accessible alternative to monkey esophagus.52, 53 Anti-RA are detected by immunofluorescence on rat tissue. The result is considered positive only when the characteristic staining pattern is seen on all three liver, kidney and stomach rodent tissues. There are at least 5 anti-reticulin staining patterns but only the type 1 (R1-anti RA) is associated with celiac disease. However, these antibodies are not used in routine practice. The discovery of anti-tTG as the autoantigen recognized by the endomysial antibodies is seen as a major breakthrough.54 This led to the development of ELISA based assays that detect anti-tissue transglutaminase antibodies (anti-tTGA). Initially anti­ tTGA tests used commercially available guinea pig antigen. After cloning and production of human tTG, this has now become the preferred antigen. Some laboratories have used highly sensitive human recombinant Indian Journal of Pediatrics, Volume 73—August, 2006

transglutaminase radioimmunoassay. A recent systematic review of the diagnostic performance of the serological tests reported in studies from 1996 to 2003 showed that the pooled specificity of the anti-EMA monkey esophagus or human umbilical chord was almost 100% in children. 55 The pooled specificity of the tTG-guinea pig and human recombinant antibodies were between 95% and 99%. The sensitivities of the anti-EMA monkey esophagus and human umbilical chord was 96% and 97% respectively while that of the anti tTG-guinea pig was 93% and human recombinant 96%. However, there is evidence to show that the sensitivities of these assays is less than 90% when milder histological grades are used as gold standards.55 It must be noted that these data have been obtained largely from studies conducted in a research setting and it is possible that the tests will be less accurate when used in a clinical setting. Also the positive predictive values will decrease when applied in the general population. While the performance of the anti-EMA is superior to anti-tTGA,55 interpretation of immunofluorescence assay (IFA) for anti-EMA tests is operator-dependent and more liable to errors in less experienced hands. ELISA based detection of anti-tTGA is a simpler and more accessible diagnostic tool and anti-human recombinant transglutaminase antibodies is reported to perform better to the anti guinea pig transglutaminase antibodies.56 The disadvantage of the anti EMA and anti-tTGA is that they are unreliable in less than 2 years.33,57 Further they are IgA dependent antibodies and will be negative in those with selective IgA deficiency, as is seen in 3% of individuals with celiac disease.14 In children with a high clinical and histological suspicion of celiac disease but a negative IgA based serology, IgA deficiency needs to be excluded. The IgG AGA or the more recently evaluated IgG EMA or the IgG tTGA would be useful in these situations. However, the IgG tTGA has produced many false positive tests and needs to be evaluated further.58 Preliminary results with tTG IgG1 subgroup antibodies have also been encouraging and could be an useful diagnostic tool for IgA deficient celiac disease. 59 Prevalence of celiac disease in type I diabetes increased from 6.4 with IgA EMA to 13.8% when IgGI EMA was used for screening. 60 Anti-tTGA would be particularly useful for screening those who are at genetic risk for celiac disease like Down’s syndrome, Type 1 diabetes and first degree relatives. However, there are certain limitations. Despite the high sensitivity, the positive predictive value of a positive anti­ tTGA test in asymptomatic genetically at risk children was reported as 67% to 84%, much lower than that reported in symptomatic children.61,62 Therefore, while screening of individuals, at risk, a positive anti-tTGA may identify a high proportion of normal intestinal biopsies. Since the positive cut-offs have been designed to optimize the sensitivity of the test, it is recommended that using higher cut-off values while using these tests for screening 705

68

Shinjini Bhatnagar and Nitya Tandon asymptomatic individuals at risk will improve the specificity and the positive predictive value. Barker et al have shown that the sensitivity and specificity was as high as 98% and 97% respectively when cut-off values of > 100 U and