the serologic diagnosis of celiac disease in adults...dr. p. vermeersch dr. g. mariën and prof. dr....
TRANSCRIPT
Critically Appraised Topic
The serologic diagnosis of celiacThe serologic diagnosis of celiac
Critically Appraised Topic
The serologic diagnosis of celiac The serologic diagnosis of celiac disease in adultsdisease in adults
Dr. P. VermeerschDr. G. Mariën and Prof. Dr. X. Bossuyt
Introduction
Celiac disease is an autoimmune disorder characterized by an- Celiac disease is an autoimmune disorder characterized by animmunologic responsiveness to ingested gluten
- Diagnosis is based on intestinal biopsy (Marsh classification)
G ti tibilit (HLA DQ2 d DQ8)- Genetic susceptibility (HLA DQ2 and DQ8)
- Prevalence in the general population estimated at 0.5-1%Prevalence in the general population estimated at 0.5 1%
- Most patients are diagnosed as adults and rarely present with overtceliac disease
Clinical Presentation
Hopper, A. D et al. BMJ 2007;335:558-562
Pathogenesis
Lumen Partially digested gluten peptides
Epithelial barrier Increased permeability of mucosa
Subepithelial region Gln GlutTG DQ2 / DQ8tTGAntibodies to
- Gliadin- Deamidated gliadin
tTG
APC
DQ2 / DQ8
T cellIFN-γ
B cell
- tTG- Neo-epitope
T-cellB-cell
Marsh Classification
Marsh 1 Marsh 2Intra-epithelial CryptMarsh 1 Marsh 2Intra-epitheliallymphocytosis
Crypthyperplasia
Marsh 3b Marsh3cSubtotal villous atrophy Total villous atrophy
Serology- Serologic testing is traditionally performed as a screening assay in
patients suspected of celiac disease
- Current methods include the detection of antibodies direct against:1. Endomysium (indirect immunofluorescence)2. Tissue transglutaminase (ELISA)3. Gliadin (ELISA)
- Detection of IgA antibodies is considered the most sensitive and
specific
- IgA anti-endomysial antibodies and IgA anti-tTG antibodies areIgA anti endomysial antibodies and IgA anti tTG antibodies are
considered the best screening assay in adults (sens. and spec. >90%).
Meta-analysis (2005, Rostom et al)
Studies in adults N Sensitivity (95%
CI)Specificity (95%
CI) Prev PPV NPVadults CI) CI)IgA-AGA 11 0.75–0.90 (H) 0.80–0.90 (H) 0.36 H H
IgG-AGA 7 0.17-1.00 (H) 0.70-0.80 (H) 0.37 H H
IgA-EMA (ME) 11 0.974 (0.957-0.985) 0.996 (0.988-0.999) 0.40 0.974 0.996
IgA-EMA (HUC) 6 0.902 (0.859-0.934) 1.000 (0.991-1.000) 0.33 0.902 1.000g ( ) ( ) ( )
IgA-tTG (GP) 5 0.859 (0.808-0.898) 0.953 (0.930-0.969) 0.31 0.859 0.953
IgA-tTG (HR) 3 0.0.981 (0.901-0.997) 0.981 (0.958-0.991) 0.16 0.981 0.981IgA tTG (HR) 3 0.0.981 (0.901 0.997) 0.981 (0.958 0.991) 0.16 0.981 0.981
H: Heterogeneous; ME: monkey oesophagus; HUC: human umbilical cord; GP: guinea pig; HR: human recombinant
Rostom et al. Gastroenterology 2005;128:S38-S46
Meta-analysis (2005, Rostom et al)
Studies in Adults N Sensitivity (95%
CI)Specificity (95%
CI) Prev PPV NPVAdults CI) CI)IgA-AGA 11 0.75–0.90 (H) 0.80–0.90 (H) 0.36 H H
IgG-AGA 7 0.17-1.00 (H) 0.70-0.80 (H) 0.37 H H
IgA-EMA (ME) 11 0.974 (0.957-0.985) 0.996 (0.988-0.999) 0.40 0.974 0.996
IgA-EMA (HUC) 6 0.902 (0.859-0.934) 1.000 (0.991-1.000) 0.33 0.902 1.000g ( ) ( ) ( )
IgA-tTG (GP) 5 0.859 (0.808-0.898) 0.953 (0.930-0.969) 0.31 0.859 0.953
IgA-tTG (HR) 3 0.0.981 (0.901-0.997) 0.981 (0.958-0.991) 0.16 0.981 0.981IgA tTG (HR) 3 0.0.981 (0.901 0.997) 0.981 (0.958 0.991) 0.16 0.981 0.981
H: Heterogenous; ME: monkey oesophagus; HUC: human umbilical cord; GP: ginea pig; HR: human recombinant
1) Performance of anti-gliadin antibodies is inferior to anti-endomysium or anti-tTG
Meta-analysis (2005, Rostom et al)
Studies in Adults N Sensitivity (95%
CI)Specificity (95%
CI) Prev PPV NPVAdults CI) CI)IgA-AGA 11 0.75–0.90 (H) 0.80–0.90 (H) 0.36 H H
IgG-AGA 7 0.17-1.00 (H) 0.70-0.80 (H) 0.37 H H
IgA-EMA (ME) 11 0.974 (0.957-0.985) 0.996 (0.988-0.999) 0.40 0.974 0.996
IgA-EMA (HUC) 6 0.902 (0.859-0.934) 1.000 (0.991-1.000) 0.33 0.902 1.000g ( ) ( ) ( )
IgA-tTG (GP) 5 0.859 (0.808-0.898) 0.953 (0.930-0.969) 0.31 0.859 0.953
IgA-tTG (HR) 3 0.0.981 (0.901-0.997) 0.981 (0.958-0.991) 0.16 0.981 0.981IgA tTG (HR) 3 0.0.981 (0.901 0.997) 0.981 (0.958 0.991) 0.16 0.981 0.981
H: Heterogenous; ME: monkey oesophagus; HUC: human umbilical cord; GP: ginea pig; HR: human recombinant
1) Performance of anti-gliadin antibodies is inferior to anti-endomysium or anti-tTG
2) Sensitivity appears lower when more patients are included with low2) Sensitivity appears lower when more patients are included with low grade histologic damage
Meta-analysis (2005, Rostom et al)
Studies in Adults N Sensitivity (95%
CI)Specificity (95%
CI) Prev PPV NPVAdults CI) CI)IgA-AGA 11 0.75–0.90 (H) 0.80–0.90 (H) 0.36 H H
IgG-AGA 7 0.17-1.00 (H) 0.70-0.80 (H) 0.37 H H
IgA-EMA (ME) 11 0.974 (0.957-0.985) 0.996 (0.988-0.999) 0.40 0.974 0.996
IgA-EMA (HUC) 6 0.902 (0.859-0.934) 1.000 (0.991-1.000) 0.33 0.902 1.000g ( ) ( ) ( )
IgA-tTG (GP) 5 0.859 (0.808-0.898) 0.953 (0.930-0.969) 0.31 0.859 0.953
IgA-tTG (HR) 3 0.0.981 (0.901-0.997) 0.981 (0.958-0.991) 0.16 0.981 0.981IgA tTG (HR) 3 0.0.981 (0.901 0.997) 0.981 (0.958 0.991) 0.16 0.981 0.981
H: Heterogenous; ME: monkey oesophagus; HUC: human umbilical cord; GP: ginea pig; HR: human recombinant
1) Performance of anti-gliadin antibodies is inferior to anti-endomysium or anti-tTG
2) Sensitivity appears lower when more patients are included with low2) Sensitivity appears lower when more patients are included with low grade histologic damage
3) PPV is likely lower when applied to a low prevalence population
IgA deficiency
- Selective IgA deficiency: primary immunodeficiency characterized by
a selective deficiency of IgA in patients with normal serum levels of
IgG and IgM in whom other causes of hypogammaglobulinemia have
been excluded.
- The frequency in Western Europe is estimated at 1/400-1/900.
- When IgA antibodies are determined it is important to rule out- When IgA antibodies are determined, it is important to rule out
selective IgA deficiency since CD occurs 10-15 times more often in
these patients than in the general populationthese patients than in the general population.
=> Determine serum IgA in all patients who have a IgA anti-tTG
Current Algorythm
Request IgA anti-tTG
Serum IgAg
Normal or increased(Adults ≥0.82 g/L)
Decreased(Adults <0.82 g/L)
IgA anti-tTGIgG anti-Gliadin
130 ti t 2050 ti t130 patients 2050 patients
Critical Appraisal
- Questions have been raised regarding the diagnostic performance of
IgA anti-tTG testing in routine clinical practice.
Th t d hi h iti iti d ifi iti i ht b l t d t- 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 specificity of the IgG anti-gliadin assay is not good- The specificity of the IgG anti-gliadin assay is not good.
- There are reports that increased serum IgA can cause false-positive
IgA anti-tTG results, but this has not systematically been studied.
Questions
1) What is the diagnostic performance of IgA anti-tTG in routine clinical
practice?p
2) Does taking into account IgA anti-tTG titer and IgA concentration
improve clinical interpretation?
3) Is the detection of IgG antibodies against deamidated gliadin3) Is the detection of IgG antibodies against deamidated gliadin
peptides (IgG DGP-AGA) a better alternative than IgG anti-gliadin
tib di (I G AGA) i ti t ith l ti I A d fi i ?antibodies (IgG AGA) in patients with a selective IgA deficiency?
Questions
1) What is the diagnostic performance of IgA anti-tTG in routine clinical
practice?p
2) Does taking into account IgA anti-tTG titer and IgA concentration
improve clinical interpretation?
3) Is the detection of IgG antibodies against deamidated gliadin3) Is the detection of IgG antibodies against deamidated gliadin
peptides (IgG DGP-AGA) a better alternative than IgG anti-gliadin
tib di (I G AGA) i ti t ith l ti I A d fi i ?antibodies (IgG AGA) in patients with a selective IgA deficiency?
Methods
- Retrospective analysis over a 42-month period to determine the
performance of IgA anti-tTG
C ti I A d fi i t ( 0 82 /L) ti t d 16- Consecutive non-IgA deficient (≥0.82 g/L) patients aged 16 years or
older who had a IgA anti-tTG and for whom biopsy results were
available were identified (558/2050).
- Patients who were previously diagnosed with celiac disease or- Patients who were previously diagnosed with celiac disease or
dermatitis herpetiformis, a severe skin manifestation of gluten
sensitivity associated with celiac disease or that were on a glutensensitivity associated with celiac disease, or that were on a gluten-
free diet were excluded.
MethodsDiagnosis of celiac disease by the clinician was consideredconfirmed when:- Duodenal biopsy showed Marsh 3
- Duodenal biopsy showed Marsh 1 or 2 and the patient responded to a glutenfree diet clinically or on duodenal biopsy
Ski bi i di d d i i h if i d h l i di d- Skin biopsy indicated dermatitis herpetiformis and the lesions disappearedwith a gluten free diet
Patient was diagnosed as non-celiac disease when:- Duodenal biopsy showed Marsh 0 and the clinician did not consider the
biopsy to be false-negative
Results
Celiac disease Non-celiac diseaseDemographic data
Number of patients 48 558
M l /F l 15/33 207/351Male/Female 15/33 207/351
Duodenal biopsyDuodenal biopsyMarsh 0 0 522
Marsh 1 7 35
Marsh 2 3 1#
Marsh 3 34 0
D titi h tif i 4 0Dermatitis herpetiformis 4 0
# P ti t di d ith i di i# Patient was diagnosed with giardiasis
ResultsPatient results
IgA anti-tTG CD Non CD
<7 U/mL 2 513≥7 U/mL 46 45
Total 48 558
ResultsPatient results
IgA anti-tTG CD Non CD
<7 U/mL 2 513≥7 U/mL 46 45
Total 48 558
O erall sensiti it and specificitOverall sensitivity and specificity
IgA anti-tTG CD Non CD
<7 U/mL 0.04 0.92≥7 U/mL 0.96 0.08
ResultsLikelihood ratio
IgA anti-tTG CD Non CD LR
<7 U/mL 0.04 0.92 0.04≥7 U/mL 0.96 0.08 11.9
LR Interpretation1 No clinical value
>10<0 1 Clinically important differences in pretest-posttest probability<0.12-5
0.2-0.5 Small difference, may be relevant in certain clinical settings
5-100.1-0.2 Modest, but substantial difference in pretest-posttest probability
ResultsLikelihood ratio
IgA anti-tTG CD Non CD LR
<7 U/mL 0.04 0.92 0.04≥7 U/mL 0.96 0.08 11.9
BUT: Positive predictive value was only 50.5% with a prevalence of7.9% in patients who had a biopsy
ResultsLikelihood ratio
IgA anti-tTG CD Non CD LR
<7 U/mL 0.04 0.92 0.04≥7 U/mL 0.96 0.08 11.9
BUT: Positive predictive value was only 50.5% with a prevalence of7.9% in patients who had a biopsy
The negative predictive value, in contrast, was 99.6%.
Summary (1)
1. Overall sensitivity (95.8%) and specificity (91.9%) of our secondgeneration IgA anti-tTG assay were good and comparable to otherstudies.
2 Given a prevalence of 7 9% in patients who had a IgA anti-tTG test2. Given a prevalence of 7.9% in patients who had a IgA anti tTG testand a biopsy, the positive predictive value (PPV) of a positive IgAanti-tTG result was only 50.5%.y
Summary (1)
1. Overall sensitivity (95.8%) and specificity (91.9%) of our secondgeneration IgA anti-tTG assay were good and comparable to otherstudies.
2 Given a prevalence of 7 9% in patients who had a IgA anti-tTG test2. Given a prevalence of 7.9% in patients who had a IgA anti tTG testand a biopsy, the positive predictive value (PPV) of a positive IgAanti-tTG result was only 50.5%.y
Of note: Since only patients with biopsy results were included, specificityis most likely underestimated and the PPV overestimated dueto a selection bias.
Selection Bias
- Clinicians are more likely to propose an intestinal biopsy in patientswith a high likelihood of celiac disease or who test false-positive.
- When IgA anti-tTG+ patients who were clinically diagnosed as non-CD and IgA anti-tTG- patients who did not require a biopsy wereCD and IgA anti tTG patients who did not require a biopsy wereincluded as true negative:
prevalence decreases from 7 9% to 2 3%- prevalence decreases from 7.9% to 2.3%- PPV decreases from 50.5% to 40.9%- Specificity increases from 91 9% to 97 8%Specificity increases from 91.9% to 97.8%
Questions
1) What is the diagnostic performance of IgA anti-tTG in routine clinical
practice?p
2) Does taking into account IgA anti-tTG titer and IgA concentration
improve clinical interpretation?
3) Is the detection of IgG antibodies against deamidated gliadin3) Is the detection of IgG antibodies against deamidated gliadin
peptides (IgG DGP-AGA) a better alternative than IgG anti-gliadin
tib di (I G AGA) i ti t ith l ti I A d fi i ?antibodies (IgG AGA) in patients with a selective IgA deficiency?
Results
IgA anti-tTG
80
100
60
80
40
20
CD + DH no CD0
CD Non CD
IgA anti-tTG concentration
IgA anti-tTG concentration
Patientswith biopsy
All patients
IgA anti-tTG concentration
Marsh 1 Marsh 2 Marsh 3Patients
with biopsy
ResultsLikelihood ratio
IgA patients LH+ CD LH+ non CD LR
0.82-2.00 g/L 247 0.929 0.034 27.32.00-4.53 g/L 325 0.960 0.097 9.9
>4.53 g/L 34 1.000 0.32 3.1
- IgA anti-tTG only modestly increases pretest-posttest probability in patientswith an increased IgA concentrationwith an increased IgA concentration
- Increased IgA is associated with significantly higher percentage of false-
iti lt (32% 6 9%)positive results (32% vs. 6.9%)
- BUT: prevalence of celiac disease was also higher in patients with an
increased IgA concentration (26% versus 6.8%)
Likelihood ratiosLikelihood ratio
Summary (2)
Taking into account IgA anti-tTG concentration and serum IgAconcentration improves clinical interpretation.
Questions
1) What is the diagnostic performance of IgA anti-tTG in routine clinical
practice?p
2) Does taking into account IgA anti-tTG titer and IgA concentration
improve clinical interpretation?
3) Is the detection of IgG antibodies against deamidated gliadin3) Is the detection of IgG antibodies against deamidated gliadin
peptides (IgG DGP-AGA) a better alternative than IgG anti-gliadin
tib di (I G AGA) i ti t ith l ti I A d fi i ?antibodies (IgG AGA) in patients with a selective IgA deficiency?
IgG anti-deamidated gliadin
alen
ce
od itivi
ty
ifici
ty
Prev
a
Test
Met
ho
Sens
i
Spec
i
Volta et al., 2008,Selected CD (M3) and diseasedcontrol patients
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
EurospitalInova
EurospitalInova
96.8%83.6%73.4%84.4%
91.0%90.3%76.9%98.5%
Villalta et al., 2007Consecutive patients with completeIgA deficiency and 113 controls
NA IgG anti-tTGIgG AGA
IgG DGP-AGA
InovaRadimInova
95%40%80%
99%87%98%
Ni l i t l 2007Niveloni et al., 2007Unselected consecutive patientsattending small bowel clinic
43% IgA anti-tTGIgA DGP-AGAIgG DGP-AGA
InovaInovaInova
95.0%98.3%96.7%
97.5%93.8%100%
Ankelo et al 2007Ankelo et al., 2007Selected CD patients and healthycontrols
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
BiofileIn-house
BiofileIn-house
90%92%78%75%
90%90%64%98%IgG DGP AGA In house 75% 98%
IgG anti-deamidated gliadin
alen
ce
od itivi
ty
ifici
ty
Prev
a
Test
Met
ho
Sens
i
Spec
i
Volta et al., 2008,Selected CD (M3) and diseasedcontrol patients
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
EurospitalInova
EurospitalInova
96.8%83.6%73.4%84.4%
91.0%90.3%76.9%98.5%
Villalta et al., 2007Consecutive patients with completeIgA deficiency and 113 controls
NA IgG anti-tTGIgG AGA
IgG DGP-AGA
InovaRadimInova
95%40%80%
99%87%98%
Ni l i t l 2007Niveloni et al., 2007Unselected consecutive patientsattending small bowel clinic
43% IgA anti-tTGIgA DGP-AGAIgG DGP-AGA
InovaInovaInova
95.0%98.3%96.7%
97.5%93.8%100%
Ankelo et al 2007Ankelo et al., 2007Selected CD patients and healthycontrols
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
BiofileIn-house
BiofileIn-house
90%92%78%75%
90%90%64%98%IgG DGP AGA In house 75% 98%
IgG anti-deamidated gliadin
alen
ce
od itivi
ty
ifici
ty
Prev
a
Test
Met
ho
Sens
i
Spec
i
Volta et al., 2008,Selected CD (M3) and diseasedcontrol patients
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
EurospitalInova
EurospitalInova
96.8%83.6%73.4%84.4%
91.0%90.3%76.9%98.5%
Villalta et al., 2007Consecutive patients with completeIgA deficiency and 113 controls
NA IgG anti-tTGIgG AGA
IgG DGP-AGA
InovaRadimInova
95%40%80%
99%87%98%
Ni l i t l 2007Niveloni et al., 2007Unselected consecutive patientsattending small bowel clinic
43% IgA anti-tTGIgA DGP-AGAIgG DGP-AGA
InovaInovaInova
95.0%98.3%96.7%
97.5%93.8%100%
Ankelo et al 2007Ankelo et al., 2007Selected CD patients and healthycontrols
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
BiofileIn-house
BiofileIn-house
90%92%78%75%
90%90%64%98%
=> Detection of IgG DGP-AGA is both more sensitive and more specific than detection of IgG AGA
IgG DGP AGA In house 75% 98%
IgG anti-deamidated gliadin
alen
ce
od itivi
ty
ifici
ty
Prev
a
Test
Met
ho
Sens
i
Spec
i
Volta et al., 2008,Selected CD (M3) and diseasedcontrol patients
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
EurospitalInova
EurospitalInova
96.8%83.6%73.4%84.4%
91.0%90.3%76.9%98.5%
Villalta et al., 2007Consecutive patients with completeIgA deficiency and 113 controls
NA IgG anti-tTGIgG AGA
IgG DGP-AGA
InovaRadimInova
95%40%80%
99%87%98%
Ni l i t l 2007Niveloni et al., 2007Unselected consecutive patientsattending small bowel clinic
43% IgA anti-tTGIgA DGP-AGAIgG DGP-AGA
InovaInovaInova
95.0%98.3%96.7%
97.5%93.8%100%
Ankelo et al 2007Ankelo et al., 2007Selected CD patients and healthycontrols
NA IgA anti-tTGIgA DGP-AGA
IgG AGAIgG DGP-AGA
BiofileIn-house
BiofileIn-house
90%92%78%75%
90%90%64%98%
=> The sensitivity and specificity of IgG DGP-AGA was not superior to the much better characterized IgA anti-tTG
IgG DGP AGA In house 75% 98%
Summary (3)
1. IgG DGP-AGA appears to be a better alternative than IgG AGA inpatients with selective IgA deficiency, but cannot replace IgA anti-tTG as routine screening assay in non-IgA deficient patients.
2 Further research is needed to determine the diagnostic2. Further research is needed to determine the diagnosticperformance, especially the false-positive rate, of IgG DGP-AGA inroutine clinical practice.p
To Do’s
1. Inform clinicians that the performance of the current IgA anti-tTGassay is good.
2. To report the likelihood ratio of the relevant interval to clinicians(including confidence intervals)(including confidence intervals).
3. Evaluate the 2 currently available IgG DGP-AGA assays (Inova andEuroimmune) on the consecutive non-IgA deficient patients.
4 Replace IgG AGA with IgG DGP AGA in patients with selective IgA4. Replace IgG AGA with IgG DGP-AGA in patients with selective IgAdeficiency. The cost of both assays is comparable.
Acknowledgements
Dep. of Laboratory Medicine (UZ Leuven)p y ( )Prof. Dr. X. BossuytDr. G. MariënApr. D. Coenen
Dep of Gastroenterology (UZ Leuven)Dep. of Gastroenterology (UZ Leuven)Prof. Dr. M. Hiele
Dep. of Pathology (UZ Leuven)Prof. Dr. K. Geboes