Antibody levels in adult patients with coeliac disease during gluten-free diet: a rapid initial decrease of clinical importance


Magnus Ström, Clinic of Endocrinology and Gastroenterology, Linköping University Hospital, s-581 85 Linköping, Sweden.
(fax: +46 (0)13 223896; e-mail:


Objective.  Analysis of antibodies against tissue transglutaminase (tTG) has been shown valuable in the diagnosis of coeliac disease (CD) but how quickly serum titres decrease after introduction of a gluten-free diet (GFD) is not known in adults. CD is a well-recognized disorder amongst the general population and many persons try a GFD for fairly vague symptoms before they seek medical advice. Therefore, it is important to determine the time that the serologic tests remain predictive of the disease after the introduction of a GFD.

Methods.  Sera were taken from 22 consecutively biopsy-proven adult patients with CD in connection with the diagnostic biopsy. The patients were followed for 1 year and sera were taken after 1, 3, 6 and 12 months after start of a GFD. Sera were stored at −20 °C and analysed for IgA antibodies against gliadin, endomysium and two different commercial tTG assays based on recombinant human tTG (tTGrh) and guinea-pig liver (tTGgp).

Results.  Twenty patients could be followed during GFD and all antibody titres fell sharply within 1 month after introduction of a GFD and continued to decline during the survey interval. Thirty days after beginning the diet only 58, 84, 74 and 53% of all patients had positive antibody levels of tTGrh, tTGgp, EmA and AGA respectively.

Conclusions. As the antibodies used to confirm the diagnosis of CD fall rapidly and continue to decline following the introduction of a GFD, it is important that health care providers carefully inquire about the possibility of self-prescribed diets before patients sought medical attention.


Coeliac disease (CD) was initially regarded as a rare disorder predominantly affecting children. During the last decades the prevalence figures have risen and the disease is commonly diagnosed in adults [1–3]. Increasing awareness of complications from CD such as osteopenia [4, 5] and infertility [6] has prompted the need of reliable screening tests [7–10].

Immunoglobulin A (IgA) antibodies against gliadin (AGA) have been used for screening of CD since the 80s [7, 11, 12]. The diagnostic sensitivity, of about 75% in adults [13] is too low to be useful and some authors have reported the specificity to be only 20% [14, 15]. Antibodies against endomysium (EmA) have a higher sensitivity, about 85–98% [16–19] and a very good specificity [17, 20, 21]. However, the interpretation of the endomysial immunoflourescence pattern is subjective and oesophagal tissue from monkeys is most commonly used.

Tissue transglutaminase (tTG) was identified as the major antigen of the EmA [22] and several enzyme-linked immunoassay techniques to detect antibodies against transglutaminase have been developed. When using guinea-pig tissue as antigen Sulkanen et al. [23] reported a sensitivity of 95% and a specificity of 94%. Recent reports indicate that the sensitivity of the tTG assays might be further improved by using tTG of human (tTGrh) instead of guinea-pig origin [24–26]. Hansson et al. [27] reported a high sensitivity and specificity in children when using tTGrh and recommended this as an alternative to EmA. Furthermore, when using tTGrh Bürgin-Wolff et al. reported a sensitivity of 96% and a specificity of 99% in a mixed population of children and adults [28]. In contrast, Leon et al. [29] compared the sensitivity and the specificity for EmA, tTGgp and tTGrh amongst children and adults and found no major differences between these tests.

Coeliac disease is now a well-known disorder amongst the general population and many individuals try a gluten-free diet (GFD) for fairly vague symptoms before they seek medical advice and before an accurate diagnosis is established. It is therefore important to know how long after the adoption of a GFD one can rely upon the currently employed serologic tests to establish an accurate diagnosis. The aim of the present study was to examine the influence on the serum levels of CD-related antibodies after 1–12 months of GFD.



Serum samples were collected from 22 (12 female, median age 62 years, range 29–86) adult CD patients consecutively diagnosed at the Linköping University Hospital. All sera were stored at −20 °C until analysis. The 22 patients were followed serologically 1, 3, 6 and 12 months after start of a GFD. The patients were on a normal gluten-containing diet at the time for the first biopsy. The GFD was commenced immediately after the clinician had informed the patients about their gluten intolerance and initially they also visited a dietician. At every revisit the dietary compliance was followed up. The number of days from diagnosis to control is shown in Table 1. All patients but two underwent repeated duodenal biopsy after 1 year of GFD.

Table 1.  Number of days from start of gluten-free diet to control blood sampling
 MedianInter quartile range
1 month3228–40
3 months9489–113
6 months186181–195
1 year394362–445


Serum samples were analysed for AGA, EmA, tTGrh and tTGgp. AGA was tested using UniCAP Gliadin IgA (Pharmacia Diagnostics, Uppsala, Sweden) and was defined as positive with a result >3 mgA L−1. Detection of EmA IgA was performed with indirect immunofluorescence using monkey oesophageal tissue as antigen (in-house assay). Tissue sections from marmoset monkey oesophagus were mounted on microscopic slides. Undiluted sera and sera diluted 1 : 25 with phosphate-buffered saline (PBS) was applied to slides, which were incubated for 30 min at room temperature. After washing with PBS the sections were covered with fluorescein conjugated rabbit-antihuman IgA (Dako, Copenhagen, Denmark) for 30 min, washed with PBS and examined by fluorescence microscopy. Positive sera were further diluted (1 : 5, 1 : 10, 1 : 25, 1 : 100, 1 : 400 and 1 : 1600). Sera positive in dilution 1 : 10 or more were defined as positive.

The tTG antibody (IgA) tests were performed with two commercial ELISA-tests: (i) Celikey®, tTGrh, IgA antibody assay (Pharmacia Diagnostics, Freiburg, Germany). It was defined by the producer as positive when >8 U mL−1, negative when <5 U mL−1, borderline between 5 and 8 U mL−1, and (ii) ImmuLisa®, tTGgp with guinea-pig-derived tTG (IMMCO, Buffalo, NY, USA). The result was defined by the producer as positive when >25 U, negative when <20 U, borderline when 20–25 U. All sera were analysed using the same batch of tTGrh test. We used the upper borderline levels as a cut off for all the analyses as recommended by the producers. The intra-assay variation was measured as CV% (CV = coefficient of variation). The values for Celikey® were 6.0, 9.3 and 9.4 for low, medium and high values respectively. Corresponding values for ImmuLisa® were 3.9, 2.6 and 3.4 respectively. The inter-assay variation measured as CV% was 13.3 for Celikey® and 6.8 for ImmuLisa®, measured with a medium-high in-house control.


Three to four biopsy specimens were obtained during upper endoscopic examination of each patient from the lower part of the duodenum descendens with standard forceps. All biopsies were fixed in a 4% buffered formaldehyde solution. Fixation, embedding and cutting were carried out according to routine methods. All specimens were scrutinized by experienced pathologists and all of the patients had at least grade IIIa changes according to the Marsh classification [30]. The study was approved by the Ethics Committee at the University Hospital of Örebro nr 577/00.


One patient had IgA deficiency and was excluded. Sixteen (76%) of the remaining 21 patients were AGA positive, 18 of 21 (86%) EmA positive, 15 of 21 (71%) tTGrh positive and 20 of 21 (95%) tTGgp positive with antibody levels above the cut off.

Of the 21 patients one stopped the GFD. The antibodies initially declined but rose after change of diet and this patient was withdrawn from further evaluation. In the other twenty patients, whose background data, weight and haemoglobin before and after 1 year of GFD are shown in Table 2, the antibody titres were continuously falling after introduction of a GFD. One month after start of a GFD 58, 84, 74 and 53% of all patients had positive antibody levels of tTGrh, tTGgp, EmA and AGA respectively (Table 3).

Table 2.  Symptoms at diagnosis, weight and haemoglobin before and after gluten-free diet for 1 year
 No. of patients
GI symptoms at diagnosis15
Diarrhoea/loose stools11
Weight loss2
 Median valuesInterquartile range
Weight before diet (kg)6863–72
Weight after diet (kg)7064–78
Haemoglobin before diet (g L−1)126116–136
Haemoglobin after diet (g L−1)129122–137
Table 3.  Patients with positive antibody tests before and after gluten-free diet for 1 year amongst patients (n = 20) with biopsy-proven coeliac disease
 0 month1 montha3 monthsa6 months12 monthsa
  1. aOne patient did not have serum samples taken. The values in parentheses are percentages.

AGA15/20 (75)10/19 (53)5/19 (26)1/20 (5)0/19 (0)
EmA17/20 (85)14/19 (74)11/19 (58)6/20 (30)2/19 (11)
tTGrh14/20 (70)11/19 (58)6/19 (32)4/20 (20)0/19 (0)
tTGgp18/20 (90)16/19 (84)12/19 (63)9/20 (45)6/19 (32)

Of the patients with positive tTGrh at the start 85, 43 and 29% were still positive after 1, 3 and 6 months respectively. After 12 months none of the patients was tTGrh or AGA positive and 13% were positive against EmA. Thirty-five per cent had tTGgp positive titres after 12 months of diet (Table 4).

Table 4.  Patients with initially positive antibody levels and biopsy-proven coeliac disease before and after a gluten-free diet for 1 year
 0 month1 month3 months6 months12 months
  1. aOne patient did not have serum samples taken. The values in parentheses are percentages.

AGA (n = 15)15/15 (100)10/14 (71)a5/14 (36)a1/15 (7)0/15 (0)
EmA (n = 17)17/17 (100)14/16 (88)a10/17 (59)6/17 (35)2/16 (13)a
tTGrh (n = 14)14/14 (100)11/13 (85)a6/14 (43)4/14 (29)0/14 (0)
tTGgp (n = 18)18/18 (100)16/17 (94)a12/17 (71)a9/18 (50)6/17 (35)a

Five patients with initially high (up to four times of cut off) tTGrh antibody titres had a lower percentage decrease (16%) in antibody levels after 1 month than nine patients with very high (more than four times of cut off) antibody titres (decrease 55%). After 3 months of diet the percentage decrease was 83 and 91% respectively and none of the five patients with initially high antibody titre compared with 67% (6/9) with initially very high antibody titres had titres over the cut off level (Fig. 1).

Figure 1.

Human recombinant tissue transglutaminase antibody levels (tTGrh) before and after 1, 3, 6 and 12 months of gluten-free diet amongst 20 patients with newly diagnosed coeliac disease. Median values amongst four groups of patients with initially normal, borderline, high (up to four times cut off) and very high (more than four times cut off) antibody levels respectively.

Of the patients who underwent a second biopsy after 1 year of GFD 16 of 18 were in remission. One had unaltered Marsh grade IIIc and negative antibodies as before GFD, one had Marsh grade IIIa and no antibody titres over cut off (only positive AGA before GFD). These two patients were both in remission on further follow-up biopsies.


The purpose of this study was to examine how antibody titres change over time after the introduction of a GFD in adults with CD. In clinical practice patients often have omitted gluten-containing food before consulting a clinician. It is essential to know if the serological tests are useful in these subjects. Several recently published studies reports on tTG and EmA in patients with CD after 1 year of GFD [28, 31, 32]. We felt it was important to compare the different CD-related serum antibodies at close intervals after the initiation of a GFD in adults with CD.

After 1 month of GFD 42% had tTGrh antibody titres below the cut off lever. By 3 months, 68% of the tTGrh titres had fallen below the cut off level. After 12 months none of our patients had antibody titres over the cut off level although three (16%) still had borderline levels. Similar results have also been found in treated children [27].

The tTGrh antibody titres decreased faster in patients with very high initial levels but the titres remained positive for a longer time than in those with lower initial serum titres. All patients with antibody titres up to maximal four times over the cut off showed normal antibody titres after 3 months of GFD.

When a patient has started a GFD before seeking medical advice, there is a real risk of misdiagnosis if the diagnosis is based on antibody titres alone. As almost half of the patients will have normal or intermediate antibody levels 1 month after introduction of a GFD, it seems of utmost importance to determine if the patient has self-initiated a diet prior to taking samples for analysis of tTGrh.

At the time of diagnosis 85 and 90% of the patients had positive EmA and tTGpg respectively, and after 1 month of diet 74 and 84% of them were still positive compared with 58 and 63% after 3 months of diet which is in line with other reports [33, 34].

There was a difference between antibodies against tTGrh or tTGgp. After one and 12 months respectively of GFD there were about 30% more patients with antibody titres over the cut of levels using tTGgp. Several reports have shown a better performance of the human tTG than the tTGgp in children [35, 36], possibly related to the purity of the tTGrh compared with the guinea-pig protein. Fabiani et al. [37] found that 16% of 123 EmA-negative adult patients had antibodies against tTGgp after 12 months GFD. A possible explanation was proposed that the guinea-pig antigen is more unspecific. Kaukinen et al. [31] found 16% tTGgp positive after 12 months in a group of 30 treated CD patients and the corresponding figure in the present study was 32%.

The rapid decrease in AGA after start of a GFD in our adult patients is similar to reports in children [34, 38]. Unfortunately there was no correlation between antibody titres and the status of the mucosa after 1 year of diet neither found in other reports [33, 39, 40].


All coeliac-related antibodies decline after introduction of a GFD. The self-initiation of such a diet before consulting a physician can therefore lead to a clinically important false-negative test result when utilizing antibodies against tTG to indicate the diagnosis of CD.

Conflict of interest statement

Authors T. Hansson and I. Dahlbom are members of the research staff of Pharmacia Diagnostics in Uppsala, Sweden, the manufacturer of the ELISA used for some of the antibody determinations described in this paper. No economical support from an entity with financial interests in the subject matter has occurred in the present study. Therefore there is no conflict of interest.