- Email: [email protected]
Use of likelihood ratios improves clinical interpretation of IgA anti-tTG antibody testing for celiac disease
Clinica Chimica Acta 411 (2010) 13–17
Contents lists available at ScienceDirect
Clinica Chimica Acta j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c l i n c h i m
Use of likelihood ratios improves clinical interpretation of IgA anti-tTG antibody testing for celiac disease Pieter Vermeersch a, Dries Coenen a, Karel Geboes b, Godelieve Mariën a, Martin Hiele c, Xavier Bossuyt a,⁎ a b c
Laboratory Medicine, Immunology, University Hospital Gasthuisberg, Catholic University of Leuven, Belgium Department of Pathology, University Hospital Gasthuisberg, Catholic University of Leuven, Belgium Department of Gastroenterology, University Hospital Gasthuisberg, Catholic University of Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 5 August 2009 Received in revised form 15 September 2009 Accepted 21 September 2009 Available online 30 September 2009 Keywords: Celiac disease Diagnosis Tissue-type transglutaminase IgA
a b s t r a c t Background: We investigated whether taking into account IgA anti-tissue transglutaminase antibody concentration (IgA anti-tTG) and total IgA concentration could improve clinical interpretation of serologic testing for celiac disease (CD). Methods: We retrospectively identified 43 consecutive newly diagnosed CD patients and 545 consecutive disease control patients who had an IgA anti-tTG request during the 42-month study period and for whom intestinal biopsy results were available. Results: Sensitivity and specificity of the IgA anti-tTG assay from Genesis was 95.3% and 92.7%, respectively, with a likelihood ratio (LR) of 12.4. The LR for CD markedly increased with increasing IgA anti-tTG concentration (from 2.0 for results between 7 and 20 U/ml up to 319 for results > 100 U/ml). The LR for CD was also higher in patients with a normal IgA concentration (0.82–4.53 g/L) compared to patients with an increased IgA concentration (15.3 vs. 3.1, respectively). These observations were confirmed with a second IgA anti-tTG assay from BioRad. Conclusion: Sensitivity of IgA anti-tTG was good. Specificity, however, was reduced when IgA anti-tTG was weak positive or when the IgA concentration was increased. Taking into account IgA anti-tTG concentration and IgA concentration improves clinical interpretation of serologic testing for CD. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Celiac disease is an auto-immune disorder characterized by a heightened immunological responsiveness to ingested gluten (from wheat, barley, or rye) in genetically susceptible individuals [1]. The disease is typically characterized by malabsorption that results from inflammatory injury of the mucosa of the small intestine. Originally considered a rare disorder in infants or children with severe malabsorption, celiac disease is now recognized as a common disorder diagnosed at all ages and affecting 0.5–1.0% of all adults [1,2]. In fact, most patients who are diagnosed with celiac disease present as adults, most commonly during their 40s [1,3]. The definitive diagnosis of celiac disease requires a small intestinal biopsy examination [4]. The detection of auto-antibodies is often used as first-line test to identify individuals who require a duodenal biopsy. For many years, the serologic diagnosis of celiac disease has been based on
Abbreviations: CD, celiac disease; CI, confidence interval; DGP, deamidated gliadin peptide; LR, likelihood ratio; ROC, receiver operating characteristic; tTG, tissue transglutaminase. ⁎ Corresponding author. Laboratory Medicine, Immunology, University Hospitals Leuven, Herestraat 49, B-3000 Leuven, Belgium. Fax: + 32 16 347042. E-mail address: [email protected] (X. Bossuyt). 0009-8981/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.cca.2009.09.030
the detection of anti-endomysial antibodies. The identification of tissue transglutaminase (tTG) as the target antigen of anti-endomysial antibodies in 1997 [5] resulted in the development of enzyme-linked immunosorbent assays for the detection of anti-tTG antibodies. The first-generation IgA anti-tTG assays, which used guinea pig liver tTG as antigen, suffered from false-positive results, especially in patients with chronic liver disease [6]. The second generation assays, which use highly purified or recombinant human tTG antigen, are reported to have excellent sensitivities and specificities [7,8]. Several authors have, however, raised questions regarding the diagnostic performance of IgA anti-tTG testing in routine clinical practice [2,9]. The reported high sensitivities and specificities might be related to the use of pre-selected groups of celiac disease patients (e.g. severe histological changes of the small bowel) and/or controls. The sensitivity of IgA anti-tTG has been reported to be lower in patients with limited mucosal damage (e.g. Marsh I and Marsh II) [10]. As a consequence, sensitivity decreases when more patients with a lesser degree of mucosal damage are included in the study. It has also been suggested, based on a limited number of observations, that specificity is reduced in patients with an increased IgA concentration [11]. There are, however, no studies that have systematically examined the potential interference by increased total IgA on IgA anti-tTG testing and diagnostic accuracy in routine clinical practice.
14
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
Traditionally, a single cutoff is used for the clinical interpretation of IgA anti-tTG results. All values above or below the cutoff are given the same interpretation (positive or negative, respectively). However, it is reasonable to assume that the likelihood of celiac disease increases with increasing IgA anti-tTG antibody concentration. Villalta et al., for example, reported that IgA anti-tTG values in patients with liver cirrhosis who tested false-positive for IgA anti-tTG were lower than IgA anti-tTG values in patients diagnosed with celiac disease [12]. This additional information is lost when results are interpreted using a single cutoff. Likelihood ratios allow to overcome this problem since they can be calculated for any given test result interval. The likelihood ratio (LR) is the ratio of the probability that a patient with the disease has a particular test result over the probability that a patient without the disease has the same test result and can be calculated for any given test result interval. An LR of 10, for example, means that a patient with the disease is 10 times more likely to have a certain test result than a patient without the disease. LRs also offer the advantage that they are independent of disease prevalence, in contrast to the positive and negative predictive value. Using LRs, one can calculate the posttest probability for any given pretest probability. We have previously shown that the use of LRs improves the clinical interpretation of enzyme immunoassays for the diagnosis of small-vessel vasculitis [13] and serum FLC testing for the diagnosis of malignant plasma cell disorders [14]. In the present paper, we examined whether taking into account IgA anti-tTG concentration and IgA concentration improves clinical interpretation of serologic testing for celiac disease. The evaluation was done on 588 different consecutive non-IgA deficient patients in whom clinicians for the first time suspected celiac disease and who underwent intestinal biopsy. Patients with known celiac disease were excluded. In all patients, IgA anti-tTG and total IgA was determined. We illustrate that the diagnostic performance of IgA anti-tTG antibodies depends (i) on the anti-tTG antibody concentration and (ii) on the total IgA concentration. 2. Materials and methods
The patient characteristics of the 43 consecutive non-IgA deficient patients diagnosed with CD and the 545 consecutive non-IgA deficient patients diagnosed as non-CD are given in Table 1. 2.2. IgA anti-tTG measurement and intestinal biopsies Serum IgA was measured by immunonephelometry on an Immage nephelometer (Beckman Coulter) to exclude selective serum IgA deficiency (serum IgA < 0.82 g/L). Routine IgA anti-tTG antibody concentrations were determined using a commercially available ELISA (Genesis, Cambridgeshire, United Kingdom) with human recombinant tTG as antigen. The ELISA was performed according to the manufacturer's instructions and was carried out on an automated ELISA instrument (PhD System, BioRad). Results were expressed as arbitrary U/ml. The recommended cutoff value by the manufacturer (≥7 U/ml) was used. To confirm the observations with the IgA anti-tTG assay from Genesis, the sera from the 588 patients included in the retrospective analysis were tested with a second IgA anti-tTG assay (BioRad, Hercules, CA) which contains tTG crosslinked to gliadin-specific peptides. These tTG neo-epitopes would be structurally more closely related to the physiological antigen [15]. Patient sera were stored at −20 °C. The recommended cutoff value by the manufacturer (>15U/ml) was used. Intestinal biopsies were taken in the second or third part of the duodenum and graded according to a modified Marsh classification [16]. Intestinal biopsies were graded by a pathologist with more than 30 years experience (KG). The pathologist was unaware of the IgA anti-tTG result. 2.3. Statistical analysis Pre- and posttest probability were calculated as described by Moore and Weatherford [17]. A Fisher's exact test was used to analyse 2 × 2 contingency tables and calculations were performed using GraphPad QuickCalcs (GraphPad Software). Confidence intervals (CI) for positive likelihood ratios (LR) were calculated with the method described by Simel et al. [18]. ROC plot analysis was calculated using Analyse-it version 2.12 (Analyse-it Software Ltd., Leeds, UK).
2.1. Study population After receiving approval from the Institutional Ethics Committee of the University Hospital of the University of Leuven, all patients aged 16 years or older attending the University Hospitals of Leuven who had a serum IgA anti-tTG antibody test at the request of the treating physician between May 1st 2004 and October 31st 2007 (42 months) were identified retrospectively. Only patients for whom biopsy results were available were included for further analysis. For evaluation of the diagnostic performance, only the first test result was included for each patient and all patients who were previously diagnosed with celiac disease or who were on a gluten-free diet were excluded. Medical records of all patients were checked. The diagnosis of celiac disease was based on a combination of characteristic, though not specific, changes on duodenal biopsy and the clinical presentation. The diagnosis of celiac disease (CD) was considered confirmed in patients with Marsh IIIa to IIIc lesions on intestinal biopsies and in patients with Marsh I or II lesions on intestinal biopsies who responded to a gluten-free diet serologically or on intestinal biopsy. The diagnosis of non-CD was considered confirmed when intestinal biopsy showed a Marsh 0 at initial presentation or when intestinal biopsy showed Marsh I or Marsh II and the morphologic lesion could be explained by another disease such as Helicobacter pylori gastritis or giardiasis according to the physician. Patients who were classified as false-positive were tested for the presence of IgG anti-deamidated gliadin peptide (DGP) antibodies with a commercially available IgG anti-DGP assay (Inova diagnostics (San Diego, CA). The clinician was not aware of the IgG anti-DGP result.
Table 1 Characteristics of the patients diagnosed as celiac disease (CD) and non-celiac disease (non-CD). CD
Non-CD
Demographic data Number of patients Male/female, n Mean age (years)
43 12/31 43.7
545 199/346 39.8
Biopsy results Marsh 0 Marsh I Marsh II Marsh III(a–c)
0 7 3 33
510 34 1a 0
Genesis IgA anti-tTG result < 7 U/ml 7–20 U/ml > 20–100 U/ml > 100 U/ml
2 6 10 25
505 38 1 1
BioRad IgA anti-tTG result ≤ 15 U/ml > 15–40 U/ml > 40–100 U/ml > 150 U/ml
6 5 3 29
518 19 6 2
IgA concentration 0.82–4.53 g/L > 4.53 g/L
36 7
520 25
a
This patient was diagnosed with giardiasis.
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
15
3. Results 3.1. Sensitivity and specificity of IgA anti-tTG Overall sensitivity and specificity in routine clinical practice of the Genesis IgA anti-tTG assay was 95.3% and 92.7%, respectively, with a likelihood ratio (LR) of 12.4 (CI 9.2–16.7). The cutoff of 7 U/ml was confirmed as the optimum cutoff on the basis of the ROC curve (maximizing sensitivity and specificity). Given a prevalence of 7.3% in our study population, the positive predictive value was only 50.6%. Of the 43 patients diagnosed with CD, 41 had an IgA anti-tTG result of ≥7 U/ml with Genesis. Histopathologic results in the 41 IgA anti-tTGpositive patients diagnosed with celiac disease on duodenal biopsy were as follows: seven patients were classified as Marsh I, three as Marsh II, and 31 as Marsh III(a–c): two patients with celiac disease (based on clinical presentation and duodenal biopsy) tested negative with the Genesis assay. These two patients had a Marsh III(a–c) lesion on intestinal biopsy. The IgA anti-tTG results in these two patients were 1.8 and 0.8 U/ml. The sensitivity and specificity of the BioRad assay, which was not performed in routine clinical practice, was 86.0% and 95.0%, respectively, with an LR of 17.4 (CI 11.8–25.6). The cutoff of 15 as recommended by the manufacturer was confirmed by ROC curve analysis (maximizing sensitivity and specificity). The IgA anti-tTG results in the 6 patients that were false-negative varied between 1.3 and 10.7. 3.2. Details of false-positive results with Genesis Forty patients had a positive IgA anti-tTG result with Genesis but were diagnosed as non-celiac disease by the treating physician. Thirtyseven patients were classified as Marsh 0, two patients as Marsh I, and one patient as Marsh II. One of the two patients classified as Marsh I was subsequently diagnosed with Crohn's disease, while the other patient, who had disturbed liver function tests at initial presentation, was Marsh 0 during follow-up. The patient with a Marsh II lesion suffered from giardiasis. Two patients had an IgA anti-tTG result of >20U/ml with Genesis. The first patient was diagnosed with auto-immune hepatitis (>100U/ml, Marsh 0) and the second patient had iron deficiency of unknown origin (35.4U/ml, Marsh 0). The diagnoses in the other 38 patients were functional gastro-intestinal complaints (n=10), iron or vitamin deficiency of unknown cause (n =4), Crohn's disease (n = 3), gastrooesophagal reflux (n = 2), disturbed liver function tests (n = 3), unexplained abdominal complaints (n=2), Helicobacter pylori-associated gastritis (n=2), gastric ulcer (n=1), gastro-intestinal tumor (n=2), type 1 diabetes mellitus (n=2), skin disorder (n =1), Waldenström's disease (n=1), intestinal giardiasis (n= 1), paralytic ileus (n= 1), membranous glomerulonephritis (n=1), discrete duodenitis (n=1), and burning mouth syndrome (n=1). All patients that were classified as false-positive with Genesis were negative for IgG anti-DGP antibodies. 3.4. High IgA anti-tTG results are associated with a higher LR for CD Most patients (81%) that were diagnosed with CD had an IgA antitTG result of >20, while 39 of the 42 patients who were considered false-positive had an IgA anti-tTG result between 7 and 20 U/ml, suggesting that taking into account IgA anti-tTG concentration could improve clinical interpretation. In order to examine this, we calculated the LRs for different Genesis IgA anti-tTG test result intervals. The LRs were 0.05 (CI 0.01–0.2), 2.0 (CI 0.9–4.5), 127 (CI 17–967) and 317 (CI 44–2282) for test result intervals of <7 U/ml, 7–20 U/ml, 20–100 and >100 U/ml, respectively (see Fig. 1A). These results clearly demonstrate that the LR depends on the anti-tTG concentration. IgA anti-tTG concentration should be taken into account when interpreting an IgA anti-tTG result.
Fig. 1. Likelihood and likelihood ratio for different IgA anti-tTG concentration intervals for Genesis IgA anti-tTG and BioRad IgA anti-tTG.
The posttest probabilities for the different test result intervals as a function of pretest probabilities were calculated through application of Bayes' Theorem. The results are shown in Fig. 2. This graphical approach illustrates how the posttest probability of CD depends on the pretest probability and the concentration of anti-tTG antibodies. The estimated posttest probability increases with both increasing pretest probability and increasing antibody concentration. The pretest probability for a random patient in our study population who underwent an intestinal biopsy was 7.3%. For a patient with Marsh I, Marsh II or Marsh III(a–c) on intestinal biopsy, the probability that the patient was diagnosed with CD was 17%, 75%, and 100%, respectively. Our observation that high IgA anti-tTG results are associated with a higher probability of CD was confirmed with the second IgA anti-tTG assay from BioRad. The LRs with the BioRad IgA anti-tTG assay were 0.1 (CI 0.07–0.3), 3.3 (CI 1.3–8.5), 6.3 (CI 1.6–24), and 184 (CI 45–744) for test result intervals of ≤15U/ml, >15–40U/ml, >40–100U/ml, and >100U/ml, respectively (see Fig. 1B). Since LR of the test result interval of >40–100U/ml was comparable to the test result interval of >15–40U/ml, they were combined into one test result interval of >15–100U/ml with an LR of 4.1 (CI 1.9–8.4) for further analysis. The posttest probabilities for the different test result intervals as a function of pretest probabilities calculated through application of Bayes' Theorem are shown in Fig. 3. 3.5. Specificity is reduced in patients with an increased total IgA concentration To examine whether an increased IgA concentration affects the specificity of IgA anti-tTG testing, we determined the LR for different total IgA concentration intervals. With Genesis, specificity was 93.5% for a normal IgA concentration (0.82–4.53 g/L) and the LR was 15.3 (CI 10.9–21.7). When the total IgA concentration exceeded 4.53 g/L,
16
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
Fig. 2. Posttest probability for celiac disease for different pretest probabilities for Genesis IgA anti-tTG. The upper and lower panels represent pretest probabilities between 0 and 0.1 and 0 and 1, respectively. The probability that a patient with Marsh I, II or III was diagnosed with CD in our study is indicated in the lower panel.
Fig. 3. Posttest probability for celiac disease for different pretest probabilities for BioRad IgA anti-tTG. The upper and lower panels represent pretest probabilities between 0 and 0.1 and 0 and 1, respectively. The probability that a patient with Marsh I, II or III(a–c) was diagnosed with CD in our study is indicated in the lower panel.
specificity decreased to 69.3% and the LR decreased to 3.1 (CI 1.8– 5.5). With BioRad, specificity was 96.5% for a normal IgA concentration (0.82–4.53 g/L) and the LR was 23.7 (CI 14.9–38.8). When the total IgA concentration exceeded 4.53 g/L, specificity decreased to 64.0% and the LR decreased to 2.8 (CI 1.6–4.7). The reduced specificity in patients with an increased IgA concentration was associated with an increased percentage of false-positive results. With Genesis, 8 of the 25 non-CD patients with an increased total IgA concentration tested false-positive, compared to 34 of the 520 non-CD patients with a normal IgA concentration (32.0% vs. 6.5%, respectively, p < 0.01). With BioRad, 9 of the 25 non-CD patients with an increased IgA concentration tested false-positive, compared to 18 of the 520 patients with a normal IgA concentration (36.0% vs. 3.5%, respectively, p < 0.01). Five patients with an increased total IgA concentration were false-positive with both assays. The prevalence of CD, however, was also higher in patients with an increased total IgA concentration compared to patients with a normal IgA concentration (21.9% versus 6.5%, respectively, p < 0.01).
With Genesis, the LRs are lower in patients with an increased IgA concentration compared to patients with a normal IgA concentration for an IgA anti-tTG result of 7–20 and >20–100 U/ml. A strong positive IgA anti-tTG result (>100 U/ml), in contrast, almost always indicated CD, irrespective of the IgA concentration (LR ≥ 289). With the BioRad assay, the LRs were lower in patients with an increased IgA concentration compared to patients with a normal IgA concentration for an IgA anti-tTG result >15–100 U/ml and >100 U/ml. A strong positive IgA anti-tTG result with BioRad almost always indicated CD when the IgA concentration was normal (LR 346), but not when the IgA concentration was increased (LR 17).
3.6. Combination of total serum IgA and IgA anti-tTG concentration improves diagnostic accuracy To determine whether taking into account both total IgA concentration and IgA anti-tTG concentration could improve diagnostic accuracy, we calculated the LRs for different total IgA and IgA anti-tTG test result intervals. The results are displayed in Table 2.
Table 2 Likelihood ratios for different total IgA and IgA anti-tTG test result intervals. Total IgA 0.82–4.53 g/L
>4.53 g/L
Genesis IgA anti-tTG result < 7 U/ml 7–20 U/ml > 20–100 U/ml > 100 U/ml
0.06 2.8 +∞ 289
0 0 7.1 +∞
BioRad IgA anti-tTG result ≤ 15 U/ml > 15–100 U/ml > 150 U/ml
0.17 5.1 347
0 0.89 17.9
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
4. Discussion In the present paper, we evaluated the clinical interpretation of IgA anti-tTG for the diagnosis of CD in routine clinical practice. We demonstrated that one could improve the diagnostic interpretation by taking IgA anti-tTG concentration and total IgA concentration into account. The sensitivity (95.3%) and specificity (92.7%) of the Genesis IgA anti-tTG assay used in our study was good and comparable to previous studies [8,19,20]. However, given a prevalence of 7.3% of CD in our study population, the positive predictive value of a positive IgA antitTG result was only 50.6%. Specificity was reduced in individuals with slightly elevated IgA anti-tTG concentrations (7–20 U/ml) and in individuals in whom the IgA concentration was increased. The sensitivity (86.0%) and specificity (95.0%) of the BioRad IgA anti-tTG assay was also good. The performance of both assays, however, cannot be directly compared since the results of the Genesis assay were available to physicians which might have caused a bias towards higher sensitivity and lower specificity for the Genesis assay. Traditionally, a single cutoff is used for the interpretation of IgA anti-tTG results and all values above or below the cutoff value are given the same interpretation (positive or negative, respectively). In routine clinical practice, however, the probability of CD is largely dependent on the pretest probability and the IgA anti-tTG concentration. Several authors have previously demonstrated that higher IgA anti-tTG antibody titers are associated with higher positive predictive values for CD [21,22]. Based on these results, it has been suggested that a biopsy might not be mandatory in patients an IgA anti-tTG antibody titer above a certain assay-specific cutoff. Such an approach does not, however, take into account the pretest probability. In the present study, we provide likelihood ratios for test result intervals which allow to calculate the posttest probability for any given pretest probability. The results clearly demonstrate that the probability of CD for any given pretest probability is largely dependent on the IgA anti-tTG concentration and on the total IgA concentration. A graphical representation of pretest and posttest probabilities could be provided to clinicians together with the test results in order to help the clinician to better interpret a given test result. Several possible limitations of this study need to be considered. First, some patients that were classified as false-positive because the duodenal biopsy result was Marsh 0 could suffer from silent or latent CD without histological features. Examination of patient records indicated that none of the 40 patients that were considered falsepositive were diagnosed with celiac disease at a later time point. All 40 patients were also negative for IgG anti-DGP antibodies, suggesting that they were indeed false-positive. Second, we excluded patients for whom no biopsy results were available which might have caused a selection bias for the Genesis assay since clinicians will tend to perform an intestinal biopsy in all patients that test positive, even if the likelihood of CD is low. We therefore performed a separate analysis including all patients who had an IgA anti-tTG request during the 42-month study period. For this analysis, we considered the 1418 patients who were IgA anti-tTG negative and did not have an intestinal biopsy and were not known with CD to be true negative and the 25 patients who were clinically diagnosed as non-CD without the need of a duodenal biopsy (IgA anti-tTG 7.0–11.5 U/ml) to be false-
17
positive. This analysis gave similar results with regard to the effect of IgA anti-tTG concentration and IgA concentration on the LR (data not shown). In conclusion, we demonstrated how LRs and posttest probability for CD depend on the IgA anti-tTG concentration. Sensitivity of IgA anti-tTG was very good, but specificity was reduced when anti-tTG was weakly positive or when the IgA concentration was increased. Taking into account the IgA anti-tTG concentration and the IgA concentration improves clinical interpretation of serologic testing for CD.
References [1] Hopper AD, Hadjivassiliou M, Butt S, Sanders DS. Adult coeliac disease. BMJ 2007;335:558–62. [2] Green PH, Cellier C. Celiac disease. N Engl J Med 2007;357:1731–43. [3] Green PHR, Stavropoulos SN, Panagi SG, et al. Characteristics of adult celiac disease in the USA: results of a national survey. Am J Gastroenterol 2001;96:126–31. [4] Dewar DH, Ciclitira PJ. Clinical features and diagnosis of celiac disease. Gastroenterology 2005;128:S19–24. [5] Dieterich W, Ehnis T, Bauer M, et al. Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat Med 1997;3:797–801. [6] Carroccio A, Giannitrapani L, Soresi M, et al. Guinea pig transglutaminase immunolinked assay does not predict coeliac disease in patients with chronic liver disease. Gut 2001;49:506–11. [7] Rostom A, Dube C, Cranney A, et al. The diagnostic accuracy of serologic tests for celiac disease: a systematic review. Gastroenterology 2005;128:S38–46. [8] Van Meensel B, Hiele M, Hoffman I, et al. Diagnostic accuracy of ten secondgeneration (human) tissue transglutaminase antibody assays in celiac disease. Clin Chem 2004;50:2125–35. [9] Abrams JA, Brar P, Diamond B, Rotterdam H, Green PH. Utility in clinical practice of immunoglobulin a anti-tissue transglutaminase antibody for the diagnosis of celiac disease. Clin Gastroenterol Hepatol 2006;4:726–30. [10] Tursi A, Brandimarte G, Giorgetti GM. Prevalence of antitissue transglutaminase antibodies in different degrees of intestinal damage in celiac disease. J Clin Gastroenterol 2003;36:219–21. [11] Hill PG, McMillan SA. Anti-tissue transglutaminase antibodies and their role in the investigation of coeliac disease. Ann Clin Biochem 2006;43:105–17. [12] Villalta D, Crovatto M, Stella S, Tonutti E, Tozzoli R, Bizzaro N. False positive reactions for IgA and IgG anti-tissue transglutaminase antibodies in liver cirrhosis are common and method-dependent. Clin Chim Acta 2005;356:102–9. [13] Vermeersch P, Blockmans D, Bossuyt X. Use of likelihood ratios can improve the clinical usefulness of enzyme immunoassays for the diagnosis of small vessel vasculitides. Clin Chem 2009;55:1886–8. [14] Vermeersch P, Vercammen M, Holvoet A, Vande Broeck I, Delforge M, Bossuyt X. Use of interval-specific likelihood ratios improves clinical interpretation of serum FLC results for the diagnosis of malignant plasma cell disorders. Clin Chim Acta in press [Electronic publication ahead of print]. [15] Ciccocioppo R, Di Sabatino A, Ara C, et al. Gliadin and tissue transglutaminase complexes in normal and coeliac duodenal mucosa. Clin Exp Immunol 2003;134:516–24. [16] Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologists. Eur J Gastroenterol Hepatol 1999;11:1185–94. [17] Moore JH, Weatherford LR. Decision modeling with Microsoft Excel. Upper Saddle River, New Jersey: Prentice-Hall, Inc; 2001. [18] Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol 1991;44:763–70. [19] Hill PG, Forsyth JM, Semeraro D, Holmes GK. IgA antibodies to human tissue transglutaminase: audit of routine practice confirms high diagnostic accuracy. Scand J Gastroenterol 2004;39:1078–82. [20] Zintzaras E, Germenis AE. Performance of antibodies against tissue transglutaminase for the diagnosis of celiac disease: meta-analysis. Clin Vaccine Immunol 2006;13:187–92. [21] Barker CC, Mitton C, Jevon G, Mock T. Can tissue transglutaminase antibody titers replace small-bowel biopsy to diagnose celiac disease in select pediatric populations? Pediatrics 2005;115:1341–6. [22] Hill PG, Holmes GK. Coeliac disease: a biopsy is not always necessary for diagnosis. Aliment Pharmacol Ther 2008;27:572–7.
Contents lists available at ScienceDirect
Clinica Chimica Acta j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c l i n c h i m
Use of likelihood ratios improves clinical interpretation of IgA anti-tTG antibody testing for celiac disease Pieter Vermeersch a, Dries Coenen a, Karel Geboes b, Godelieve Mariën a, Martin Hiele c, Xavier Bossuyt a,⁎ a b c
Laboratory Medicine, Immunology, University Hospital Gasthuisberg, Catholic University of Leuven, Belgium Department of Pathology, University Hospital Gasthuisberg, Catholic University of Leuven, Belgium Department of Gastroenterology, University Hospital Gasthuisberg, Catholic University of Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 5 August 2009 Received in revised form 15 September 2009 Accepted 21 September 2009 Available online 30 September 2009 Keywords: Celiac disease Diagnosis Tissue-type transglutaminase IgA
a b s t r a c t Background: We investigated whether taking into account IgA anti-tissue transglutaminase antibody concentration (IgA anti-tTG) and total IgA concentration could improve clinical interpretation of serologic testing for celiac disease (CD). Methods: We retrospectively identified 43 consecutive newly diagnosed CD patients and 545 consecutive disease control patients who had an IgA anti-tTG request during the 42-month study period and for whom intestinal biopsy results were available. Results: Sensitivity and specificity of the IgA anti-tTG assay from Genesis was 95.3% and 92.7%, respectively, with a likelihood ratio (LR) of 12.4. The LR for CD markedly increased with increasing IgA anti-tTG concentration (from 2.0 for results between 7 and 20 U/ml up to 319 for results > 100 U/ml). The LR for CD was also higher in patients with a normal IgA concentration (0.82–4.53 g/L) compared to patients with an increased IgA concentration (15.3 vs. 3.1, respectively). These observations were confirmed with a second IgA anti-tTG assay from BioRad. Conclusion: Sensitivity of IgA anti-tTG was good. Specificity, however, was reduced when IgA anti-tTG was weak positive or when the IgA concentration was increased. Taking into account IgA anti-tTG concentration and IgA concentration improves clinical interpretation of serologic testing for CD. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Celiac disease is an auto-immune disorder characterized by a heightened immunological responsiveness to ingested gluten (from wheat, barley, or rye) in genetically susceptible individuals [1]. The disease is typically characterized by malabsorption that results from inflammatory injury of the mucosa of the small intestine. Originally considered a rare disorder in infants or children with severe malabsorption, celiac disease is now recognized as a common disorder diagnosed at all ages and affecting 0.5–1.0% of all adults [1,2]. In fact, most patients who are diagnosed with celiac disease present as adults, most commonly during their 40s [1,3]. The definitive diagnosis of celiac disease requires a small intestinal biopsy examination [4]. The detection of auto-antibodies is often used as first-line test to identify individuals who require a duodenal biopsy. For many years, the serologic diagnosis of celiac disease has been based on
Abbreviations: CD, celiac disease; CI, confidence interval; DGP, deamidated gliadin peptide; LR, likelihood ratio; ROC, receiver operating characteristic; tTG, tissue transglutaminase. ⁎ Corresponding author. Laboratory Medicine, Immunology, University Hospitals Leuven, Herestraat 49, B-3000 Leuven, Belgium. Fax: + 32 16 347042. E-mail address: [email protected] (X. Bossuyt). 0009-8981/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.cca.2009.09.030
the detection of anti-endomysial antibodies. The identification of tissue transglutaminase (tTG) as the target antigen of anti-endomysial antibodies in 1997 [5] resulted in the development of enzyme-linked immunosorbent assays for the detection of anti-tTG antibodies. The first-generation IgA anti-tTG assays, which used guinea pig liver tTG as antigen, suffered from false-positive results, especially in patients with chronic liver disease [6]. The second generation assays, which use highly purified or recombinant human tTG antigen, are reported to have excellent sensitivities and specificities [7,8]. Several authors have, however, raised questions regarding the diagnostic performance of IgA anti-tTG testing in routine clinical practice [2,9]. The reported high sensitivities and specificities might be related to the use of pre-selected groups of celiac disease patients (e.g. severe histological changes of the small bowel) and/or controls. The sensitivity of IgA anti-tTG has been reported to be lower in patients with limited mucosal damage (e.g. Marsh I and Marsh II) [10]. As a consequence, sensitivity decreases when more patients with a lesser degree of mucosal damage are included in the study. It has also been suggested, based on a limited number of observations, that specificity is reduced in patients with an increased IgA concentration [11]. There are, however, no studies that have systematically examined the potential interference by increased total IgA on IgA anti-tTG testing and diagnostic accuracy in routine clinical practice.
14
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
Traditionally, a single cutoff is used for the clinical interpretation of IgA anti-tTG results. All values above or below the cutoff are given the same interpretation (positive or negative, respectively). However, it is reasonable to assume that the likelihood of celiac disease increases with increasing IgA anti-tTG antibody concentration. Villalta et al., for example, reported that IgA anti-tTG values in patients with liver cirrhosis who tested false-positive for IgA anti-tTG were lower than IgA anti-tTG values in patients diagnosed with celiac disease [12]. This additional information is lost when results are interpreted using a single cutoff. Likelihood ratios allow to overcome this problem since they can be calculated for any given test result interval. The likelihood ratio (LR) is the ratio of the probability that a patient with the disease has a particular test result over the probability that a patient without the disease has the same test result and can be calculated for any given test result interval. An LR of 10, for example, means that a patient with the disease is 10 times more likely to have a certain test result than a patient without the disease. LRs also offer the advantage that they are independent of disease prevalence, in contrast to the positive and negative predictive value. Using LRs, one can calculate the posttest probability for any given pretest probability. We have previously shown that the use of LRs improves the clinical interpretation of enzyme immunoassays for the diagnosis of small-vessel vasculitis [13] and serum FLC testing for the diagnosis of malignant plasma cell disorders [14]. In the present paper, we examined whether taking into account IgA anti-tTG concentration and IgA concentration improves clinical interpretation of serologic testing for celiac disease. The evaluation was done on 588 different consecutive non-IgA deficient patients in whom clinicians for the first time suspected celiac disease and who underwent intestinal biopsy. Patients with known celiac disease were excluded. In all patients, IgA anti-tTG and total IgA was determined. We illustrate that the diagnostic performance of IgA anti-tTG antibodies depends (i) on the anti-tTG antibody concentration and (ii) on the total IgA concentration. 2. Materials and methods
The patient characteristics of the 43 consecutive non-IgA deficient patients diagnosed with CD and the 545 consecutive non-IgA deficient patients diagnosed as non-CD are given in Table 1. 2.2. IgA anti-tTG measurement and intestinal biopsies Serum IgA was measured by immunonephelometry on an Immage nephelometer (Beckman Coulter) to exclude selective serum IgA deficiency (serum IgA < 0.82 g/L). Routine IgA anti-tTG antibody concentrations were determined using a commercially available ELISA (Genesis, Cambridgeshire, United Kingdom) with human recombinant tTG as antigen. The ELISA was performed according to the manufacturer's instructions and was carried out on an automated ELISA instrument (PhD System, BioRad). Results were expressed as arbitrary U/ml. The recommended cutoff value by the manufacturer (≥7 U/ml) was used. To confirm the observations with the IgA anti-tTG assay from Genesis, the sera from the 588 patients included in the retrospective analysis were tested with a second IgA anti-tTG assay (BioRad, Hercules, CA) which contains tTG crosslinked to gliadin-specific peptides. These tTG neo-epitopes would be structurally more closely related to the physiological antigen [15]. Patient sera were stored at −20 °C. The recommended cutoff value by the manufacturer (>15U/ml) was used. Intestinal biopsies were taken in the second or third part of the duodenum and graded according to a modified Marsh classification [16]. Intestinal biopsies were graded by a pathologist with more than 30 years experience (KG). The pathologist was unaware of the IgA anti-tTG result. 2.3. Statistical analysis Pre- and posttest probability were calculated as described by Moore and Weatherford [17]. A Fisher's exact test was used to analyse 2 × 2 contingency tables and calculations were performed using GraphPad QuickCalcs (GraphPad Software). Confidence intervals (CI) for positive likelihood ratios (LR) were calculated with the method described by Simel et al. [18]. ROC plot analysis was calculated using Analyse-it version 2.12 (Analyse-it Software Ltd., Leeds, UK).
2.1. Study population After receiving approval from the Institutional Ethics Committee of the University Hospital of the University of Leuven, all patients aged 16 years or older attending the University Hospitals of Leuven who had a serum IgA anti-tTG antibody test at the request of the treating physician between May 1st 2004 and October 31st 2007 (42 months) were identified retrospectively. Only patients for whom biopsy results were available were included for further analysis. For evaluation of the diagnostic performance, only the first test result was included for each patient and all patients who were previously diagnosed with celiac disease or who were on a gluten-free diet were excluded. Medical records of all patients were checked. The diagnosis of celiac disease was based on a combination of characteristic, though not specific, changes on duodenal biopsy and the clinical presentation. The diagnosis of celiac disease (CD) was considered confirmed in patients with Marsh IIIa to IIIc lesions on intestinal biopsies and in patients with Marsh I or II lesions on intestinal biopsies who responded to a gluten-free diet serologically or on intestinal biopsy. The diagnosis of non-CD was considered confirmed when intestinal biopsy showed a Marsh 0 at initial presentation or when intestinal biopsy showed Marsh I or Marsh II and the morphologic lesion could be explained by another disease such as Helicobacter pylori gastritis or giardiasis according to the physician. Patients who were classified as false-positive were tested for the presence of IgG anti-deamidated gliadin peptide (DGP) antibodies with a commercially available IgG anti-DGP assay (Inova diagnostics (San Diego, CA). The clinician was not aware of the IgG anti-DGP result.
Table 1 Characteristics of the patients diagnosed as celiac disease (CD) and non-celiac disease (non-CD). CD
Non-CD
Demographic data Number of patients Male/female, n Mean age (years)
43 12/31 43.7
545 199/346 39.8
Biopsy results Marsh 0 Marsh I Marsh II Marsh III(a–c)
0 7 3 33
510 34 1a 0
Genesis IgA anti-tTG result < 7 U/ml 7–20 U/ml > 20–100 U/ml > 100 U/ml
2 6 10 25
505 38 1 1
BioRad IgA anti-tTG result ≤ 15 U/ml > 15–40 U/ml > 40–100 U/ml > 150 U/ml
6 5 3 29
518 19 6 2
IgA concentration 0.82–4.53 g/L > 4.53 g/L
36 7
520 25
a
This patient was diagnosed with giardiasis.
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
15
3. Results 3.1. Sensitivity and specificity of IgA anti-tTG Overall sensitivity and specificity in routine clinical practice of the Genesis IgA anti-tTG assay was 95.3% and 92.7%, respectively, with a likelihood ratio (LR) of 12.4 (CI 9.2–16.7). The cutoff of 7 U/ml was confirmed as the optimum cutoff on the basis of the ROC curve (maximizing sensitivity and specificity). Given a prevalence of 7.3% in our study population, the positive predictive value was only 50.6%. Of the 43 patients diagnosed with CD, 41 had an IgA anti-tTG result of ≥7 U/ml with Genesis. Histopathologic results in the 41 IgA anti-tTGpositive patients diagnosed with celiac disease on duodenal biopsy were as follows: seven patients were classified as Marsh I, three as Marsh II, and 31 as Marsh III(a–c): two patients with celiac disease (based on clinical presentation and duodenal biopsy) tested negative with the Genesis assay. These two patients had a Marsh III(a–c) lesion on intestinal biopsy. The IgA anti-tTG results in these two patients were 1.8 and 0.8 U/ml. The sensitivity and specificity of the BioRad assay, which was not performed in routine clinical practice, was 86.0% and 95.0%, respectively, with an LR of 17.4 (CI 11.8–25.6). The cutoff of 15 as recommended by the manufacturer was confirmed by ROC curve analysis (maximizing sensitivity and specificity). The IgA anti-tTG results in the 6 patients that were false-negative varied between 1.3 and 10.7. 3.2. Details of false-positive results with Genesis Forty patients had a positive IgA anti-tTG result with Genesis but were diagnosed as non-celiac disease by the treating physician. Thirtyseven patients were classified as Marsh 0, two patients as Marsh I, and one patient as Marsh II. One of the two patients classified as Marsh I was subsequently diagnosed with Crohn's disease, while the other patient, who had disturbed liver function tests at initial presentation, was Marsh 0 during follow-up. The patient with a Marsh II lesion suffered from giardiasis. Two patients had an IgA anti-tTG result of >20U/ml with Genesis. The first patient was diagnosed with auto-immune hepatitis (>100U/ml, Marsh 0) and the second patient had iron deficiency of unknown origin (35.4U/ml, Marsh 0). The diagnoses in the other 38 patients were functional gastro-intestinal complaints (n=10), iron or vitamin deficiency of unknown cause (n =4), Crohn's disease (n = 3), gastrooesophagal reflux (n = 2), disturbed liver function tests (n = 3), unexplained abdominal complaints (n=2), Helicobacter pylori-associated gastritis (n=2), gastric ulcer (n=1), gastro-intestinal tumor (n=2), type 1 diabetes mellitus (n=2), skin disorder (n =1), Waldenström's disease (n=1), intestinal giardiasis (n= 1), paralytic ileus (n= 1), membranous glomerulonephritis (n=1), discrete duodenitis (n=1), and burning mouth syndrome (n=1). All patients that were classified as false-positive with Genesis were negative for IgG anti-DGP antibodies. 3.4. High IgA anti-tTG results are associated with a higher LR for CD Most patients (81%) that were diagnosed with CD had an IgA antitTG result of >20, while 39 of the 42 patients who were considered false-positive had an IgA anti-tTG result between 7 and 20 U/ml, suggesting that taking into account IgA anti-tTG concentration could improve clinical interpretation. In order to examine this, we calculated the LRs for different Genesis IgA anti-tTG test result intervals. The LRs were 0.05 (CI 0.01–0.2), 2.0 (CI 0.9–4.5), 127 (CI 17–967) and 317 (CI 44–2282) for test result intervals of <7 U/ml, 7–20 U/ml, 20–100 and >100 U/ml, respectively (see Fig. 1A). These results clearly demonstrate that the LR depends on the anti-tTG concentration. IgA anti-tTG concentration should be taken into account when interpreting an IgA anti-tTG result.
Fig. 1. Likelihood and likelihood ratio for different IgA anti-tTG concentration intervals for Genesis IgA anti-tTG and BioRad IgA anti-tTG.
The posttest probabilities for the different test result intervals as a function of pretest probabilities were calculated through application of Bayes' Theorem. The results are shown in Fig. 2. This graphical approach illustrates how the posttest probability of CD depends on the pretest probability and the concentration of anti-tTG antibodies. The estimated posttest probability increases with both increasing pretest probability and increasing antibody concentration. The pretest probability for a random patient in our study population who underwent an intestinal biopsy was 7.3%. For a patient with Marsh I, Marsh II or Marsh III(a–c) on intestinal biopsy, the probability that the patient was diagnosed with CD was 17%, 75%, and 100%, respectively. Our observation that high IgA anti-tTG results are associated with a higher probability of CD was confirmed with the second IgA anti-tTG assay from BioRad. The LRs with the BioRad IgA anti-tTG assay were 0.1 (CI 0.07–0.3), 3.3 (CI 1.3–8.5), 6.3 (CI 1.6–24), and 184 (CI 45–744) for test result intervals of ≤15U/ml, >15–40U/ml, >40–100U/ml, and >100U/ml, respectively (see Fig. 1B). Since LR of the test result interval of >40–100U/ml was comparable to the test result interval of >15–40U/ml, they were combined into one test result interval of >15–100U/ml with an LR of 4.1 (CI 1.9–8.4) for further analysis. The posttest probabilities for the different test result intervals as a function of pretest probabilities calculated through application of Bayes' Theorem are shown in Fig. 3. 3.5. Specificity is reduced in patients with an increased total IgA concentration To examine whether an increased IgA concentration affects the specificity of IgA anti-tTG testing, we determined the LR for different total IgA concentration intervals. With Genesis, specificity was 93.5% for a normal IgA concentration (0.82–4.53 g/L) and the LR was 15.3 (CI 10.9–21.7). When the total IgA concentration exceeded 4.53 g/L,
16
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
Fig. 2. Posttest probability for celiac disease for different pretest probabilities for Genesis IgA anti-tTG. The upper and lower panels represent pretest probabilities between 0 and 0.1 and 0 and 1, respectively. The probability that a patient with Marsh I, II or III was diagnosed with CD in our study is indicated in the lower panel.
Fig. 3. Posttest probability for celiac disease for different pretest probabilities for BioRad IgA anti-tTG. The upper and lower panels represent pretest probabilities between 0 and 0.1 and 0 and 1, respectively. The probability that a patient with Marsh I, II or III(a–c) was diagnosed with CD in our study is indicated in the lower panel.
specificity decreased to 69.3% and the LR decreased to 3.1 (CI 1.8– 5.5). With BioRad, specificity was 96.5% for a normal IgA concentration (0.82–4.53 g/L) and the LR was 23.7 (CI 14.9–38.8). When the total IgA concentration exceeded 4.53 g/L, specificity decreased to 64.0% and the LR decreased to 2.8 (CI 1.6–4.7). The reduced specificity in patients with an increased IgA concentration was associated with an increased percentage of false-positive results. With Genesis, 8 of the 25 non-CD patients with an increased total IgA concentration tested false-positive, compared to 34 of the 520 non-CD patients with a normal IgA concentration (32.0% vs. 6.5%, respectively, p < 0.01). With BioRad, 9 of the 25 non-CD patients with an increased IgA concentration tested false-positive, compared to 18 of the 520 patients with a normal IgA concentration (36.0% vs. 3.5%, respectively, p < 0.01). Five patients with an increased total IgA concentration were false-positive with both assays. The prevalence of CD, however, was also higher in patients with an increased total IgA concentration compared to patients with a normal IgA concentration (21.9% versus 6.5%, respectively, p < 0.01).
With Genesis, the LRs are lower in patients with an increased IgA concentration compared to patients with a normal IgA concentration for an IgA anti-tTG result of 7–20 and >20–100 U/ml. A strong positive IgA anti-tTG result (>100 U/ml), in contrast, almost always indicated CD, irrespective of the IgA concentration (LR ≥ 289). With the BioRad assay, the LRs were lower in patients with an increased IgA concentration compared to patients with a normal IgA concentration for an IgA anti-tTG result >15–100 U/ml and >100 U/ml. A strong positive IgA anti-tTG result with BioRad almost always indicated CD when the IgA concentration was normal (LR 346), but not when the IgA concentration was increased (LR 17).
3.6. Combination of total serum IgA and IgA anti-tTG concentration improves diagnostic accuracy To determine whether taking into account both total IgA concentration and IgA anti-tTG concentration could improve diagnostic accuracy, we calculated the LRs for different total IgA and IgA anti-tTG test result intervals. The results are displayed in Table 2.
Table 2 Likelihood ratios for different total IgA and IgA anti-tTG test result intervals. Total IgA 0.82–4.53 g/L
>4.53 g/L
Genesis IgA anti-tTG result < 7 U/ml 7–20 U/ml > 20–100 U/ml > 100 U/ml
0.06 2.8 +∞ 289
0 0 7.1 +∞
BioRad IgA anti-tTG result ≤ 15 U/ml > 15–100 U/ml > 150 U/ml
0.17 5.1 347
0 0.89 17.9
P. Vermeersch et al. / Clinica Chimica Acta 411 (2010) 13–17
4. Discussion In the present paper, we evaluated the clinical interpretation of IgA anti-tTG for the diagnosis of CD in routine clinical practice. We demonstrated that one could improve the diagnostic interpretation by taking IgA anti-tTG concentration and total IgA concentration into account. The sensitivity (95.3%) and specificity (92.7%) of the Genesis IgA anti-tTG assay used in our study was good and comparable to previous studies [8,19,20]. However, given a prevalence of 7.3% of CD in our study population, the positive predictive value of a positive IgA antitTG result was only 50.6%. Specificity was reduced in individuals with slightly elevated IgA anti-tTG concentrations (7–20 U/ml) and in individuals in whom the IgA concentration was increased. The sensitivity (86.0%) and specificity (95.0%) of the BioRad IgA anti-tTG assay was also good. The performance of both assays, however, cannot be directly compared since the results of the Genesis assay were available to physicians which might have caused a bias towards higher sensitivity and lower specificity for the Genesis assay. Traditionally, a single cutoff is used for the interpretation of IgA anti-tTG results and all values above or below the cutoff value are given the same interpretation (positive or negative, respectively). In routine clinical practice, however, the probability of CD is largely dependent on the pretest probability and the IgA anti-tTG concentration. Several authors have previously demonstrated that higher IgA anti-tTG antibody titers are associated with higher positive predictive values for CD [21,22]. Based on these results, it has been suggested that a biopsy might not be mandatory in patients an IgA anti-tTG antibody titer above a certain assay-specific cutoff. Such an approach does not, however, take into account the pretest probability. In the present study, we provide likelihood ratios for test result intervals which allow to calculate the posttest probability for any given pretest probability. The results clearly demonstrate that the probability of CD for any given pretest probability is largely dependent on the IgA anti-tTG concentration and on the total IgA concentration. A graphical representation of pretest and posttest probabilities could be provided to clinicians together with the test results in order to help the clinician to better interpret a given test result. Several possible limitations of this study need to be considered. First, some patients that were classified as false-positive because the duodenal biopsy result was Marsh 0 could suffer from silent or latent CD without histological features. Examination of patient records indicated that none of the 40 patients that were considered falsepositive were diagnosed with celiac disease at a later time point. All 40 patients were also negative for IgG anti-DGP antibodies, suggesting that they were indeed false-positive. Second, we excluded patients for whom no biopsy results were available which might have caused a selection bias for the Genesis assay since clinicians will tend to perform an intestinal biopsy in all patients that test positive, even if the likelihood of CD is low. We therefore performed a separate analysis including all patients who had an IgA anti-tTG request during the 42-month study period. For this analysis, we considered the 1418 patients who were IgA anti-tTG negative and did not have an intestinal biopsy and were not known with CD to be true negative and the 25 patients who were clinically diagnosed as non-CD without the need of a duodenal biopsy (IgA anti-tTG 7.0–11.5 U/ml) to be false-
17
positive. This analysis gave similar results with regard to the effect of IgA anti-tTG concentration and IgA concentration on the LR (data not shown). In conclusion, we demonstrated how LRs and posttest probability for CD depend on the IgA anti-tTG concentration. Sensitivity of IgA anti-tTG was very good, but specificity was reduced when anti-tTG was weakly positive or when the IgA concentration was increased. Taking into account the IgA anti-tTG concentration and the IgA concentration improves clinical interpretation of serologic testing for CD.
References [1] Hopper AD, Hadjivassiliou M, Butt S, Sanders DS. Adult coeliac disease. BMJ 2007;335:558–62. [2] Green PH, Cellier C. Celiac disease. N Engl J Med 2007;357:1731–43. [3] Green PHR, Stavropoulos SN, Panagi SG, et al. Characteristics of adult celiac disease in the USA: results of a national survey. Am J Gastroenterol 2001;96:126–31. [4] Dewar DH, Ciclitira PJ. Clinical features and diagnosis of celiac disease. Gastroenterology 2005;128:S19–24. [5] Dieterich W, Ehnis T, Bauer M, et al. Identification of tissue transglutaminase as the autoantigen of celiac disease. Nat Med 1997;3:797–801. [6] Carroccio A, Giannitrapani L, Soresi M, et al. Guinea pig transglutaminase immunolinked assay does not predict coeliac disease in patients with chronic liver disease. Gut 2001;49:506–11. [7] Rostom A, Dube C, Cranney A, et al. The diagnostic accuracy of serologic tests for celiac disease: a systematic review. Gastroenterology 2005;128:S38–46. [8] Van Meensel B, Hiele M, Hoffman I, et al. Diagnostic accuracy of ten secondgeneration (human) tissue transglutaminase antibody assays in celiac disease. Clin Chem 2004;50:2125–35. [9] Abrams JA, Brar P, Diamond B, Rotterdam H, Green PH. Utility in clinical practice of immunoglobulin a anti-tissue transglutaminase antibody for the diagnosis of celiac disease. Clin Gastroenterol Hepatol 2006;4:726–30. [10] Tursi A, Brandimarte G, Giorgetti GM. Prevalence of antitissue transglutaminase antibodies in different degrees of intestinal damage in celiac disease. J Clin Gastroenterol 2003;36:219–21. [11] Hill PG, McMillan SA. Anti-tissue transglutaminase antibodies and their role in the investigation of coeliac disease. Ann Clin Biochem 2006;43:105–17. [12] Villalta D, Crovatto M, Stella S, Tonutti E, Tozzoli R, Bizzaro N. False positive reactions for IgA and IgG anti-tissue transglutaminase antibodies in liver cirrhosis are common and method-dependent. Clin Chim Acta 2005;356:102–9. [13] Vermeersch P, Blockmans D, Bossuyt X. Use of likelihood ratios can improve the clinical usefulness of enzyme immunoassays for the diagnosis of small vessel vasculitides. Clin Chem 2009;55:1886–8. [14] Vermeersch P, Vercammen M, Holvoet A, Vande Broeck I, Delforge M, Bossuyt X. Use of interval-specific likelihood ratios improves clinical interpretation of serum FLC results for the diagnosis of malignant plasma cell disorders. Clin Chim Acta in press [Electronic publication ahead of print]. [15] Ciccocioppo R, Di Sabatino A, Ara C, et al. Gliadin and tissue transglutaminase complexes in normal and coeliac duodenal mucosa. Clin Exp Immunol 2003;134:516–24. [16] Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologists. Eur J Gastroenterol Hepatol 1999;11:1185–94. [17] Moore JH, Weatherford LR. Decision modeling with Microsoft Excel. Upper Saddle River, New Jersey: Prentice-Hall, Inc; 2001. [18] Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol 1991;44:763–70. [19] Hill PG, Forsyth JM, Semeraro D, Holmes GK. IgA antibodies to human tissue transglutaminase: audit of routine practice confirms high diagnostic accuracy. Scand J Gastroenterol 2004;39:1078–82. [20] Zintzaras E, Germenis AE. Performance of antibodies against tissue transglutaminase for the diagnosis of celiac disease: meta-analysis. Clin Vaccine Immunol 2006;13:187–92. [21] Barker CC, Mitton C, Jevon G, Mock T. Can tissue transglutaminase antibody titers replace small-bowel biopsy to diagnose celiac disease in select pediatric populations? Pediatrics 2005;115:1341–6. [22] Hill PG, Holmes GK. Coeliac disease: a biopsy is not always necessary for diagnosis. Aliment Pharmacol Ther 2008;27:572–7.