Article 1

see related editorial on page x

Office-Based Point of Care Testing (IgA/IgG-Deamidated

COLON/SMALL BOWEL
Gliadin Peptide) for Celiac Disease
Michelle S. Lau, MBChB1, Peter D. Mooney, MD1, William L. White, MBChB1, Michael A. Rees, BMedSci1, Simon H. Wong, MBChB1,
Marios Hadjivassiliou, FRCP2, Peter H. R. Green, MD3, Benjamin Lebwohl, MD3 and David S. Sanders, FRCP1

Objectives: Celiac disease (CD) is common yet under-detected. A point of care test (POCT) may improve CD
detection. We aimed to assess the diagnostic performance of an IgA/IgG-deamidated gliadin peptide
(DGP)-based POCT for CD detection, patient acceptability, and inter-observer variability of the POCT
results.

Methods: From 2013–2017, we prospectively recruited patients referred to secondary care with
gastrointestinal symptoms, anemia and/or weight loss (group 1); and patients with self-reported
gluten sensitivity with unknown CD status (group 2). All patients had concurrent POCT, IgA-tissue
transglutaminase (IgA-TTG), IgA-endomysial antibodies (IgA-EMA), total IgA levels, and duodenal
biopsies. Five hundred patients completed acceptability questionnaires, and inter-observer variability
of the POCT results was compared among five clinical staff for 400 cases.

Results: Group 1: 1000 patients, 58.5% female, age 16–91, median age 57. Forty-one patients (4.1%)
were diagnosed with CD. The sensitivities of the POCT, IgA-TTG, and IgA-EMA were 82.9, 78.1,
and 70.7%; the specificities were 85.4, 96.3, and 99.8%. Group 2: 61 patients, 83% female; age
17–73, median age 35. The POCT had 100% sensitivity and negative predictive value in detecting
CD in group 2. Most patients preferred the POCT to venepuncture (90.4% vs. 2.8%). There was good
inter-observer agreement on the POCT results with a Fleiss Kappa coefficient of 0.895.

Conclusions: The POCT had comparable sensitivities to serology, and correctly identified all CD cases in a gluten
sensitive cohort. However, its low specificity may increase unnecessary investigations. Despite its
advantage of convenience and rapid results, it may not add significant value to case finding in an
office-based setting.
Am J Gastroenterol https://doi.org/10.1038/s41395-018-0143-3

Introduction NCGS using a double blind placebo controlled challenge [15], self-
Celiac disease (CD) is a systemic autoimmune disease associated reported gluten sensitivity describes individuals who complain of
with gastrointestinal and extra-gastrointestinal symptoms, trig- gastrointestinal and/or non gastrointestinal symptoms triggered by
gered by gluten in genetically susceptible individuals [1]. It affects gluten ingestion and present to physicians accordingly. Exclusion
0.3–2.4% of the general population globally [2–9]. CD affects one of CD and wheat allergy is fundamental in this group of patients. It
in 100 in the United Kingdom, but only 24% are detected [10]. is essential to distinguish NCGS from CD, as patients with NCGS
Similar observations are also apparent in Europe [3], the United do not seem to be at risk of the complications seen in CD, although
States [11], and worldwide [12]. This is partly because symptoms they derive symptomatic benefit from a gluten free diet [16]. More-
of CD can be non-specific and difficult for clinicians to recog- over, any delays in celiac testing before individuals embark on a
nize. This is further compounded by an emerging clinical entity, self-imposed gluten free diet could cause diagnostic challenges.
non-celiac gluten sensitivity (NCGS), which is clinically indistin- Early diagnosis of CD is important for the improvement of
guishable from CD [13, 14]. Although the Salerno criteria define patients’ quality of life and the prevention of complications such

1
Academic Department of Gastroenterology, Royal Hallamshire Hospital, Sheffield Teaching Hospitals, Sheffield, UK. 2Academic Department of Neurosciences
and University of Sheffield, Royal Hallamshire Hospital, Sheffield Teaching Hospitals, Sheffield, UK. 3Celiac Disease Centre, Columbia University Medical Centre,
New York, NY, USA. Correspondence: M.S.L. (email: michellelau@doctors.org.uk)
Received 26 January 2018; accepted 4 May 2018

© 2018 The American College of Gastroenterology The American Journal of Gastroenterology

 2 lau et al.

as osteoporosis, hip fractures, and lymphoproliferative malignan- Group two consisted of patients presenting to secondary care
cies. We have previously shown that serological testing in patients with self-reported gluten sensitivity, with gastrointestinal and/or
with high risk symptoms in a clinic setting yielded 3.3–4.7% CD extra-gastrointestinal symptoms related to gluten ingestion. The
detection [17]. Similar results were obtained by other groups celiac status of these patients was unknown. Patients with known
COLON/SMALL BOWEL

through case finding [18, 19]. On the other hand, a recent system- CD were excluded. Those with reduced or no gluten intake were
atic review reported insufficient evidence to support screening for asked to undertake a 6-week gluten challenge of 10 g gluten/day
asymptomatic patients at present [13, 20–23]. For these reasons, prior to their endoscopy as per guidelines [29, 30]. All patients were
case finding for CD in at risk individuals has been recommended concurrently tested with total IgA levels, IgA-TTG, IgA-EMA, and
by international guidelines [24, 25]. the POCT, and duodenal biopsies were taken in all patients.
Despite guidelines recommending celiac testing in at risk indi-
viduals, it has been shown that serological testing for CD is under- Point of care test
utilized, where only 30% of patients with suspected CD or anemia The DGP-based POCT for CD, Simtomax, was manufactured by
had serology performed prior to their endoscopy [26, 27]. This Augurix Diagnostics, Rheinfelden, Switzerland. It detects both
suggests that current case finding strategies with serology may be IgA/IgG-DGP antibodies, as well as the presence of IgA. The assay
inadequate. This could be due to a multitude of factors, including is based on lateral flow immunochromatography using colloidal
a lack of awareness of the guidelines, inconvenience and cost. A gold antihuman antibodies as a signal detector. A sample of 25 μl
finger prick point of care test (POCT) that provides convenience of capillary venous blood was obtained through a simple finger
and rapid celiac antibody results may have a role in improving case prick technique. The blood sample was then applied to the test
detection, particularly in an office-based consultation, where the device, followed by the application of five drops of the provided
results could provide immediate guidance for the physician on the buffer solution. The result was available after 10 minutes. Positive
need for duodenal biopsies. Several POCTs have been developed results were indicated by the presence of a solid red band for IgA/
in the past decade, with the majority detecting IgA-tissue trans- IgG-DGP positivity. A second single red band indicated the pres-
glutaminase (IgA-TTG). However, these POCTs have not entered ence of IgA. A third inbuilt red control band ensured a correctly
widespread clinical use, probably due to their inferior sensitivities functioning test. See Fig. 1 for illustration of the POCT.
compared to conventional serology. A recent head to head POCTs
trial comparing Biocard (IgA-TTG), Celiac Quick Test (IgA-TTG), Celiac serology
and Simtomax (IgA/IgG-deamidated gliadin peptide (DGP)) IgA-TTG was assayed using enzyme-linked immunosorbent assay
revealed that Simtomax significantly outperformed the other two, (ELISA) kits (Aesku Diagnostics, Wendelsheim, Germany). IgA-
with sensitivities of 72.2, 77.8, and 94.4%, respectively [28]. EMA was detected by immunofluorescence on primate esophagus
Our primary aim was to assess the diagnostic accuracy of sections (Binding Site, Birmingham, UK). Total IgA was meas-
the IgA/IgG-DGP-based POCT, Simtomax, in detecting CD in ured on a Behring BN2 nephelometer (Haywards Heath, West
patients presenting to secondary care with gastrointestinal symp- Sussex, UK). DGP serology was not available in our laboratory
toms, anemia and/or weight loss, and those who self-report gluten and therefore not tested.
sensitivity. Our secondary aims were to evaluate patient accept-
ability of the POCT, and the inter-observer variability of test result Histological evaluation
interpretation. In total, at least five biopsies were taken from the duodenum
with a single bite per pass technique, including at least one
biopsy from the duodenal bulb and four quadrantic biopsies
Methods from the second part of the duodenum. Each biopsy was fixed
Study design and patients in formalin at the time of the gastroscopy. Specimens were then
The study took place at the Royal Hallamshire Hospital, Sheffield, processed, orientated, and embedded in paraffin wax by the
UK, from March 2013–January 2017. We prospectively recruited pathology department. Standard 3 μm thick sections at three
patients who were referred to gastroenterology for further evalu- levels were stained with hematoxylin and eosin, and reported
ation. by gastrointestinal histopathologists without knowledge of the
Group one consisted of patients presenting to secondary care POCT or serology results. Villous atrophy was graded accord-
with gastrointestinal symptoms (abdominal pain, diarrhea and/ ing to the modified Marsh criteria [31]. The histological grade
or dyspepsia), anemia and/or weight loss. Patients with known recorded was based on the most severe grade detected from the
CD were excluded. Patients who were referred with positive celiac biopsy samples.
serology by their primary care physicians were excluded from the
study so as to avoid tertiary referral bias, thus providing a more Definitions of diagnoses
accurate assessment of the sensitivities of the POCT that is reflec- The definition of CD was based on positive serology (positive
tive of clinical practice. All patients who consented to participate TTG and/or EMA) with Marsh 3 villous atrophy.
in the study were concurrently tested with total immunoglobulin Seronegative CD was based on Marsh 3 villous atrophy on a
A (IgA) levels, IgA-TTG antibodies, IgA-endomysial antibodies normal gluten containing diet, positive human leukocyte antigen
(IgA-EMA) and the DGP-based POCT, Simtomax. An endoscopy (HLA) DQ2 or DQ8, and other supporting information such as
with duodenal biopsies was performed in all patients. family history and response to a gluten free diet. Non-celiac causes

The American Journal of Gastroenterology www.nature.com/ajg

 Point of care Test (igA/igG-DGP) for celiac Disease 3

COLON/SMALL BOWEL
Fig. 1 Three possible outcomes of the point of care test results. Red band A indicates a positive result, red band B indicates the presence of IgA, red band
CT is the control line, indicating a correctly functioning test. Left: a solid red band A indicating a positive test; Middle: an absence of a red band A indicat-
ing a negative test; Right: a faint pink band A which was classified as a negative test, as none of the patients with a faint band A had celiac disease in our
cohort

of seronegative villous atrophy were extensively investigated for, Inter-observer variability of the POCT results
including giardiasis, tuberculosis, whipple’s disease, small bowel Inter-observer variability of the POCT results was assessed
bacterial overgrowth, helicobacter pylori, human immunodefi- in 400 consecutive patients in group one. Each observer
ciency virus, autoimmune enteropathy and drug related causes. recorded whether there was a definite red band, a faint red
Marsh 3 villous atrophy secondary to CD was the reference stand- band, or an absence of a red band. There were five observ-
ard used in our study for the diagnostic performance evaluation of ers in total for each case, consisting of one gastroenterologist
the POCT and serology. and four other randomly selected allied health care profes-
Potential CD was defined as positive serology with no villous sionals (for example, nurses). All observers were trained to
atrophy (Marsh 0–2), with supporting information such as positive recognize positive, negative, and indeterminate results. Obser-
HLA DQ2 or DQ8 and family history. vation of the results was carried out indoors under fluorescent
Non-celiac gluten sensitivity was diagnosed in patients self- lighting.
reporting symptoms related to gluten who had negative serology,
absence of villous atrophy, and symptom response to a gluten free Ethical considerations
diet and gluten challenge. A 6-week gluten challenge of 10 g glu- The study protocol was approved by the Yorkshire and the
ten/day was proposed for group two patients entering the study Humber Research Ethics committee and registered with the
who had reduced or absent gluten intake prior to their investiga- local research and development department of Sheffield
tions. Teaching Hospital NHS Foundation Trust under the registra-
tion number STH15416. Written consent was obtained from all
Patient acceptability of the POCT patients.
There are no validated patient acceptability questionnaires in
the literature for POCTs. Therefore, we devised a questionnaire Statistical analysis
consisting of five questions regarding the acceptability of the Data were summarized by descriptive statistics, including counts
POCT (comfort level, convenience, and satisfaction with result and percentages for categorical data, and medians and ranges for
availability) which was filled in by 500 consecutive patients after continuous parameters. The diagnostic accuracies of the POCT,
having had the POCT performed. They were asked to rate on IgA-TTG, and IgA-EMA were presented with sensitivity, speci-
a Likert scale of one to five for each question, with one being a ficity, positive (PPV), and negative predictive values (NPV).
negative experience and five being a positive experience. These Clopper–Pearson method was used to calculate the confidence
500 patients all had previous experience of a venepuncture. They intervals for the sensitivities. Inter-observer variability was pre-
were also asked to state their preferred mode of testing: POCT, sented using Fleiss Kappa coefficient, where 0 indicates no agree-
venepuncture, or no preference. A similar acceptability question- ment and 1 indicates perfect agreement. Cohen’s effect size (r)
naire for venepuncture was completed by a separate cohort of 63 for patient acceptability between the POCT and venepuncture
patients after having had a venepuncture to act as controls. These groups was measured using Mann–Whitney U test, where r = 0.1,
questionnaires were given out to both groups to fill in indepen- 0.3, and 0.5 indicates small, medium, and large effect size, respec-
dently and anonymously, and the questionnaires were collected by tively. Statistical analysis was performed using IBM SPSS statistics
a member of staff on completion. version 24.

© 2018 The American College of Gastroenterology The American Journal of Gastroenterology

 4 lau et al.

Results patients managed a 6-week challenge and seven could only tolerate
Patient demographics and presenting characteristics in group 4 weeks of gluten challenge at which point the serology and endos-
one are illustrated in Table 1. One thousand eligible patients who copy were performed due to significant symptoms. The remaining
consented for participation entered group one of the study. There 38 patients were consuming a gluten containing diet and contin-
COLON/SMALL BOWEL

were 585/1000 females (58.5%); age range 16–91, median age 57. ued to do so at least until the investigations took place. There were
Forty-one patients (4.1%) were diagnosed with CD. IgA deficiency 51/61 females (82.9%); age range 17–73, median age 35. Eighteen
detected by total IgA levels from the laboratory assay was found patients were tested positive for EMA by their general practition-
in 29 patients in groups one and two combined (29/1061 = 2.7%). ers. The vast majority (57/61) of patients had gastrointestinal symp-
Three IgA deficient patients were diagnosed with CD (3/45 = 6.7% toms, and ten patients reported extra-gastrointestinal symptoms,
of the total celiac cohort), and all three had a positive POCT. Nine predominantly neurological complaints (e.g., headache, paresthe-
patients (9/41 = 22%) had ultra-short CD with Marsh 3 villous sia, foggy mind, ataxia, lethargy, tongue tingling, and arthralgia).
atrophy confined to the duodenal bulb only. The sensitivity of the Forty-two patients (42/61 = 68.9%) were diagnosed with NCGS,
POCT was comparable to IgA-TTG and IgA-EMA (82.9 vs. 78.1 17/61 (27.9%) with CD, and 2/61 (3.3%) with potential CD. The
vs. 70.7%). However, its specificity was significantly lower than POCT demonstrated a sensitivity and negative predictive value of
IgA-TTG and IgA-EMA (85.9 vs. 96.3 vs. 99.8%). The diagnostic 100% (vs. sensitivity 88.2%, 94.1% and negative predictive value
performance of the POCT, IgA-TTG, and IgA-EMA for group one 91.8%, 97.77% for IgA-TTG and IgA-EMA, respectively). The
are displayed in Tables  2 and 3. Receiver operating characteris- diagnostic performance of the POCT, IgA-TTG, and IgA-EMA
tic (ROC) curves for the aforementioned tests for group one are for group two are displayed in Tables 4 and 5. ROC curves for the
demonstrated in Fig. 2. aforementioned tests for group two are demonstrated in Fig. 3.
In group two, 70 patients self-reported gluten sensitivity. Nine In regards to patient acceptability, the POCT had significantly
patients who were on a self-imposed gluten free diet and declined higher patient satisfaction compared to venepuncture. The differ-
a 6-week gluten challenge prior to investigations were excluded ence in the scores between the two groups were statistically signifi-
from the study. A total of 61 patients consuming gluten entered cant in all aspects of the acceptability questionnaire, and the effect
group two of our study. Twenty-three patients who were previously size difference between the two groups was large (r = 0.506–0.656).
on a self-imposed gluten free diet underwent a gluten challenge: 16 Table 6 shows the median scores and statistical differences in both
groups for each aspect of the tests.
There was a good degree of inter-observer agreement on the
POCT result interpretation, with a Fleiss Kappa coefficient of 0.895
Table 1  Group one patient demographics and presenting
overall. Sub-analysis revealed a high level of agreement for definite
characteristics table
red bands (Kappa 0.887) and absence of red bands (Kappa 0.956).
No. of patients Celiac disease yield The level of agreement dropped for faint red bands (Kappa 0.781),
Female 585/1000 (58.5%) 27/585 (4.6%)
Male 415/1000 (41.5%) 14/415 (3.4%) Table 3  Group 1: cross tabulation of the point of care test (POCT)
Diarrhea 75/1000 (7.5%) 8/75 (10.7%) results by the reference standard
Abdominal pain 159/1000 (15.9%) 13/159 (8.2%) CD Not CD Not CD Not
Weight loss 104/1000 (10.4%) 6/104 (5.8%) CD CD CD

Anemia 194/1000 (19.4%) 9/194 (4.6%) POCT + 34 140 TTG + 32 36 EMA + 29 2

Dyspepsia 549/1000 (54.9%) 8/549 (1.5%) POCT − 7 819 TTG − 9 923 EMA − 12 957

Table 2  The diagnostic accuracy of the point of care test, IgA-tissue transglutaminase antibodies and IgA-endomysial antibodies in
detecting celiac disease in symptomatic patients (group one; n = 1000, celiac disease prevalence 4.1%)

Point of care test IgA-tissue transglutaminase antibodies IgA-endomysial antibodies

Sensitivity % (95% CI) 82.9 (67.9–92.9) 78.1 (62.4–89.4) 70.7 (54.5–83.9)

Specificity % (95% CI) 85.4 (83.0–87.6) 96.3 (94.8–97.4) 99.8 (99.3–100.0)
Positive predictive value % (95% CI) 19.5 (16.5–23.0) 47.1 (38.3–56.0) 93.6 (78.2–98.3)
Negative predictive value % (95% CI) 99.2 (98.4–99.6) 99.0 (98.3–99.5) 98.8 (98.0–99.2)
Positive likelihood ratio (95% CI) 5.7 (4.6–7.0) 20.8 (14.5–30.0) 339.2 (83.8–1373.2)
Negative likelihood ratio (95% CI) 0.2 (0.1–0.4) 0.2 (0.1–0.4) 0.3 (0.2–0.5)
Accuracy % (95% CI) 85.3 (83.0–87.4) 95.5 (94.0–96.7) 98.6 (97.7–99.2)

The American Journal of Gastroenterology www.nature.com/ajg

 Point of care Test (igA/igG-DGP) for celiac Disease 5

where there were 31 such cases within the 400 assessed. None of positive antibodies (either POCT or serology) were biopsied [32–
these 31 patients had CD. Only solid red bands were classified as a 37]. Additionally, some POCT studies measured the sensitivities
positive test for the purpose of diagnostic calculations in our study, against serology rather than duodenal histology as the reference
and faint red bands were interpreted as negative. Figure. 1 illus- standard [35, 38, 39]. These limitations could lead to a positive

COLON/SMALL BOWEL
trates the three possible outcomes of the POCTs results. ascertainment bias, thereby falsely elevating the reported sensitivi-
ties.
Another strength of this study is that our patient cohort had
Discussion a CD prevalence consistent with real life case finding in patients
To our knowledge, this is the largest study to date evaluating the
diagnostic accuracy of the DGP-based POCT, Simtomax. This
is also the first study to explore the practicalities of this POCT Table 5  Group 2: cross tabulation of the point of care test (POCT)
including patient acceptability and inter-observer variability of results by the reference standard
test result interpretation. CD Not CD Not CD Not
One of the strengths of this study is that all participants had CD CD CD
duodenal biopsies taken, irrespective of their celiac antibodies
POCT + 17 9 TTG + 15 3 EMA + 16 1
or POCT results. This ensured that no false negative cases of CD
POCT − 0 35 TTG − 2 41 EMA − 1 43
would be missed. This methodology contributed to a major dif-
ference to most POCT studies for CD, where only patients with

ROC curve ROC curve

1.0 Source of the 1.0 Source of the
curve curve
POCT POCT
EMA EMA
0.8 TTG 0.8 TTG
Reference line Reference line

0.6 0.6
Sensitivity

Sensitivity

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
1 - Specificity 1 - Specificity

Fig. 2  Group 1 receiver operating characteristic (ROC) curve for the point Fig. 3  Group 2 receiver operating characteristic (ROC) curve for the point
of care test (POCT), IgA-endomysial antibodies (EMA), and IgA-tissue of care test (POCT), IgA-endomysial antibodies (EMA) and IgA-tissue
transglutaminase antibodies (TTG). Area under the curve (AUC) for transglutaminase antibodies (TTG). Area under the curve (AUC) for each
each test were 0.842 (CI: 0.77-0.9), 0.853 (CI: 0.77-0.94), and 0.871 test were 0.898 (CI: 0.82–0.98), 0.959 (CI: 0.89–1.0) and 0.907 (CI:
(CI: 0.8-0.95), respectively. CI = confidence interval 0.81–1.0) respectively. CI = confidence interval

Table 4  The diagnostic accuracy of the point of care test, IgA-tissue transglutaminase antibodies, and IgA-endomysial antibodies in
detecting celiac disease in patients who self-reported gluten sensitivity (group two; n = 61, celiac disease prevalence 27.9%)

Point of care test IgA-tissue transglutaminase antibodies IgA-endomysial antibodies

Sensitivity % (95% CI) 100 (80.5–100) 88.2 (63.6–98.5) 94.1 (71.3–99.9)

Specificity % (95% CI) 79.6 (64.7–90.2) 93.2 (81.3–98.6) 97.7 (88.0–99.9)
Positive predictive value % (95% CI) 65.4 (51.3–77.2) 83.3 (62.3–93.8) 94.1 (69.6–99.1)
Negative predictive value % (95% CI) 100 91.8 (81.0–97.3) 97.7 (86.5–99.6)
Positive likelihood ratio (95% CI) 4.9 (2.7–8.8) 12.9 (4.3–39.1) 41.4 (5.9–288.5)
Negative likelihood ratio (95% CI) 0 0.1 (0–0.5) 0.06 (0.01–0.4)
Accuracy % (95% CI) 85.3 (73.8–93.0) 91.8 (81.9–97.3) 96.7 (88.7–99.6)

© 2018 The American College of Gastroenterology The American Journal of Gastroenterology

 6 lau et al.

Table 6  Patient acceptability for the point of care test (POCT) and conventional venepuncture. Acceptability was scored with a Likert scale
from 1 to 5, with 1 being a negative experience and 5 being a positive experience

Point of care test (POCT) Venepuncture Mann–Whitney U test

COLON/SMALL BOWEL

Blood collection process

  Score for comfort level of the test 4.7 3.3 U = 2988.5, Z = 12.027,
p < 0.001, r = 0.506
  Score for speed and ease of the test 4.7 3.3 U = 2182.5, Z = 13.443,
p < 0.001, r = 0.566
Convenience
 Satisfaction score for having the test performed during the consultation (for 4.8 2.9 U = 1086.5, Z = 14.675,
POCT) vs separately from the consultation by the phlebotomy service p < 0.001, r = 0.617
Quality of care
 Satisfaction score for obtaining test results within 10 minutes (for POCT) vs a few 4.8 3.1 U = 583.5, Z = 15.597,
days to a week (for serology via venepuncture) p < 0.001, r = 0.656
 Satisfaction score for obtaining and discussing the test results with the clinician 4.8 2.9 U = 988.0, Z = 15.223,
within the same consultation (for POCT) vs at a later date (for venepuncture) p < 0.001, r = 0.64

Preference No. of patients

Prefers point of care test 452/500 (90.4%)

Prefers venepuncture 14/500 (2.8%)
No preference 34/500 (6.8%)

with high risk symptoms, which have been reported to be 3–4.7% patients), creating positive ascertainment bias which enhanced the
[17, 18]. A much higher CD prevalence is a common limitation sensitivities of serology. In a subsequent meta-analysis of the diag-
in previous POCT studies [34, 40, 41]. This tertiary referral bias nostic accuracy of IgA-TTG and IgA-DGP [43], the authors con-
restricts the generalizability of their findings. The patient char- cluded that only two out of 11 studies biopsied controls [44, 45].
acteristics and pre-test probability of group one allowed a more In fact, these two studies demonstrated the sensitivities and spe-
accurate reflection of the diagnostic performance of these tests in cificities of IgA-TTG to be 78.3–95% and 97.5–98.4% respectively.
real practice. Second of all, the results from the aforementioned two studies still
There are a few limitations to our study. Ideally, the measurement may not have reflected their performance in real practice, as it has
of laboratory DGP serology would act as a useful comparison of been demonstrated previously [46], since the CD prevalence was
the sensitivities between DGP detection by laboratory assay (serol- very high at 74% for both studies. This again could have falsely
ogy) and lateral flow immunochromatography (POCT). However, increased the sensitivity and positive predictive value of IgA-TTG.
laboratory DGP serology is not widely available in the United Lastly, the lack of standardization of IgA-TTG laboratory assays
Kingdom (UK) and is not available in our center. Therefore, DGP could also lead to different IgA-TTG sensitivities. IgA-TTG anti-
assays were not performed. Another limitation is the evaluation body units and reference ranges are arbitrary and method-specific.
of patient acceptability of the POCT. We devised our own POCT Furthermore, over 30 different IgA-TTG assay kits are used in the
acceptability questionnaire as there were no validated question- UK, giving different IgA-TTG titers. A recent study showed that
naires in the literature, and the methodology of using a Likert scale even when the same IgA-TTG ELISA assay kit was used, there was
provided a quantitative rather than qualitative measure of accept- still poor agreement among laboratories as to whether the sam-
ability. Qualitative interviews would give a more informative rep- ple was above or below the defined IgA-TTG level cut off point
resentation of patient acceptability. However, patient acceptability for Marsh 3 histology using a ROC curve [47]. A recent head to
was a secondary outcome and not the main focus of this study. head trial of three different TTG serological kits also found widely
What is noteworthy is the generally lower sensitivities of IgA- variable sensitivities and specificities, ranging from 71.1–95.5%
TTG and IgA-EMA compared to previous serological studies [42, and 82.6–100%, respectively [48]. All these factors explain the
43]. There are several potential reasons for this. Although a system- huge variability of IgA-TTG sensitivities and why the sensitivities
atic review in 2006 showed that the pooled sensitivities of IgA-TTG appeared to be lower than average in our study, where we biopsied
and IgA-EMA from published data were 93% (range 70–100%) all patients including controls and the CD prevalence being low in
and specificities were >98% (range 90–100%) for both, the authors comparison to other studies.
indicated that these figures were likely to be falsely high due to In regards to the prevalence of CD in individuals who self-report
methodological flaws in most studies [42]. Firstly, many studies gluten sensitivity, there are four studies in the literature which
did not biopsy controls (i.e., take duodenal biopsies in seronegative assessed the diagnostic outcomes of this cohort, with sample sizes

The American Journal of Gastroenterology www.nature.com/ajg

 Point of care Test (igA/igG-DGP) for celiac Disease 7

ranging from 93 to 238, and the prevalence of CD varying between respectively). With the 100% negative predictive value, the POCT
2 and 42.4% [49–52]. In our study, the CD prevalence of 27.9% could potentially save USD $5141 per 100 endoscopies through
within the self-reported gluten sensitivity cohort lies within the duodenal biopsy avoidance (vs. routine duodenal biopsy for ane-
range of the reported data. The wide variation in the reported dis- mia) in iron deficient patients with a negative POCT when used in

COLON/SMALL BOWEL
ease prevalence is likely due to differences in the study population, the pre-endoscopy setting, unless the patient has other high-risk
study design, recruitment methods, and diagnostic criteria. For malabsorptive symptoms such as weight loss and diarrhea [27].
example, our disease prevalence of 27.9% is higher than the 7% The advantages of the POCT over conventional serology are
reported by Aziz et al. which derived from a UK population-based favorable patient acceptability and rapidly available results within
questionnaire targeting individuals with gluten related symptoms 10 minutes. Nevertheless, despite there being no significant dif-
[51], as opposed to symptomatic individuals actively presenting ference in the overall diagnostic performance between the POCT
to primary care who were then referred on to secondary care for and serology based on ROC curve analysis, one must consider the
further evaluation. Our group 2 patients’ gluten-related symptoms clinical impact of the high false positive rates of the POCT. The
may have prompted more proactive celiac screening by their gen- potential burden of a considerable increase in unnecessary inves-
eral practitioners, thus possibly explaining the higher prevalence tigations may outweigh the benefits of a sensitive and convenient
of seropositive patients (18/61), giving a higher disease prevalence. test. We have previously demonstrated that this POCT could be
Nevertheless, after excluding the 18 patients who were referred useful in CD monitoring, as it had a significantly higher sensitivity
with positive EMA in group 2, the sensitivity and negative predic- than IgA-TTG in predicting persistent villous atrophy in known
tive value of the POCT remained at 100%, where all four cases of CD patients on a gluten-free diet (67.1% vs. 47.1%, p = 0.0005)
CD were correctly identified. [56]. Although the sensitivity of the POCT was still suboptimal, it
POCTs for other laboratory measurements, such as human represented a stepwise improvement in current disease monitor-
immunodeficiency virus and international normalized ratio have ing compared to conventional serology. However, as a case find-
been widely adopted in UK practice in both primary and second- ing tool in an office-based setting where the CD prevalence would
ary care settings, owing to their clinical effectiveness and good be expected to be ~4%, as was in our study and other case find-
patient acceptability [53, 54]. We have shown that this POCT had ing studies based on symptomatic cohorts [17], the POCT may
a favorable acceptability to patients compared to venepuncture, not provide significant added value compared to conventional
with 90.4% patients preferring the POCT. Most patients generally serology due to its low specificity, albeit its similar sensitivity to
found the POCT to be a simple and quick test to perform (it took IgA-TTG.
on average 1 min to perform the test, and 10 min for the results
to become available), and less painful than venepuncture. Table 6
illustrates the satisfaction scores for different aspects of the POCT Conclusion
versus venepuncture. The DGP-based POCT had comparable sensitivities to IgA-TTG
With regards to the diagnostic performance of the POCT, the and IgA-EMA in detecting CD in symptomatic patients, and cor-
sensitivity was comparable to IgA-TTG and IgA-EMA (82.9% rectly distinguished all cases of CD in a gluten sensitive cohort that
vs. 78.1% vs. 70.7%). In our group one cohort, 7.3% (3/41) of the was consuming gluten. It also has the advantage of convenience,
newly diagnosed patients had seronegative CD detected by the rapid result availability, and good patient acceptability. However,
POCT alone whilst IgA-TTG was negative. An increase in diag- the POCT is limited by its low specificity which may increase the
nostic yield with DGP was also demonstrated by Hoerter et al. number of unnecessary investigations. The POCT therefore may
recently, where the use of IgA-DGP serology resulted in a 15% not add significant value when used for case finding in a general
increase in CD detection where IgA-TTG was negative [55]. How- office-based consultation compared to conventional serology.
ever, the specificity and PPV of the POCT were inferior to IgA-
TTG and IgA-EMA (specificities 85.4% vs. 96.3% vs. 99.8% and Conflict of interest
PPVs 19.5% vs. 47.1% vs. 93.6%, respectively), due to a higher rate Guarantor of the article: David Sanders, MBChB, MD,
of false positives. This could potentially lead to unnecessary fur- FACG, FRCP.
ther investigations. A possible explanation of the low specificity Specific author contributions: MSL is involved in data collection,
is that approximately half of the group one cohort had dyspepsia, analysis and interpretation of the data, statistical analysis, and
which constituted low risk for CD, and hence may have lowered drafting of the manuscript; PDM is involved in data collection,
the pre-test probability of CD and hence the positive predictive and analysis and interpretation of the data; WLW, MAR, SHW
value. On the other hand, when the POCT was used in higher are responsible for data collection; MH, PHRG, BL are involved
risk groups, such as patients who self-reported gluten sensitivity in analysis and interpretation of the data, and revision of the
(group two), the positive predictive value increased to 65.4%, with manuscript; DSS is responsible for study concept and design,
a 100% sensitivity and negative predictive value in detecting CD. analysis interpretation of the data, revision of the manuscript and
Similarly, we have previously shown that the POCT had better study supervision. All authors have approved the manuscript.
diagnostic performance when used in 133 patients with iron defi- Financial support: None.
ciency who were referred for an endoscopy (sensitivity, specificity, Potential competing interests: DSS has received educational
positive, and negative predictive value of 100, 82.2, 57.8, and 100%, research grants from Dr Schaer (a gluten-free food manufacturer)

© 2018 The American College of Gastroenterology The American Journal of Gastroenterology

 8 lau et al.

15. Catassi C, Elli L, Bonaz B, Bouma G, Carroccio A, Castillejo G, et al.

and Tillotts Pharma (distributor of a point of care test for celiac Diagnosis of non-celiac gluten sensitivity (NCGS): the salerno experts’
disease) for investigator led studies. Dr Schaer and Tillott’s Pharma criteria. Nutrients. 2015;7:4966–77.
did not have any input in the study design, access to study data, 16. Aziz I, Hadjivassiliou M, Sanders DS. The spectrum of noncoeliac gluten
sensitivity. Nat Rev Gastroenterol Hepatol. 2015;12:516–26.
interpretation of the findings or drafting of the manuscript.
COLON/SMALL BOWEL

17. Sanders DS, Patel D, Stephenson TJ, Ward AM, McCloskey EV,
Hadjivassiliou M, et al. A primary care cross-sectional study of undiag-
nosed adult coeliac disease. Eur J Gastroenterol Hepatol. 2003;15:407–13.
Study Highlights 18. Hin H, Bird G, Fisher P, Mahy N, Jewell D. Coeliac disease in primary
care: case finding study. BMJ. 1999;318:164–7.
19. Catassi C, Kryszak D, Louis-Jacques O, Duerksen DR, Hill I, Crowe SE,
What is current knowledge et al. Detection of Celiac disease in primary care: a multicenter case-
✓✓Celiac disease is common but underdetected, partly due to finding study in North America. Am J Gastroenterol. 2007;102:1454–60.
20. Chou R, Bougatsos C, Blazina I, Mackey K, Grusing S, Selph S. Screen-
underutilization of serology.
ing for celiac disease: evidence report and systematic review for the US
✓✓Delayed testing could cause diagnostic challenges in indi- preventive services task force. JAMA. 2017;317:1258–68.
viduals who are on a self-imposed gluten-free diet. 21. Jin J. Screening for celiac disease. JAMA. 2017;317:1286.
What is new here 22. Force USPST, Bibbins-Domingo K, Grossman DC, Curry SJ, Barry MJ,
Davidson KW, et al. Screening for celiac disease: US preventive services
✓✓The POCT had comparable sensitivities to conventional task force recommendation statement. JAMA. 2017;317:1252–7.
23. Choung RS, Murray JA. The US preventive services task force recommen-
serology, with the advantage of convenience and rapid
results. dation on screening for asymptomatic celiac disease: a dearth of evidence.
JAMA. 2017;317:1221–3.
✓✓However, the POCT may have a limited role in general 24. Ludvigsson JF, Bai JC, Biagi F, Card TR, Ciacci C, Ciclitira PJ, et al.
office-based case finding due to its low specificity. Diagnosis and management of adult coeliac disease: guidelines from the
British Society of Gastroenterology. Gut. 2014;63:1210–28.
25. Rubio-Tapia A, Hill ID, Kelly CP, Calderwood AH, Murray JA, American
College of Gastroenterology ACG clinical guidelines: diagnosis and man-
agement of celiac disease. Am J Gastroenterol. 2013;108:656–76.quiz 77.
References 26. Wiland HO, Henricks WH, Daly TM. Limited utilization of serologic
1. Lebwohl B,Sanders DS,Green PHR, Coeliac disease. Lancet. 2018;391: testing in patients undergoing duodenal biopsy for celiac disease. BMC
70–81. Gastroenterol. 2013;13:156.
2. Fasano A, Berti I, Gerarduzzi T, Not T, Colletti RB, Drago S, et al. 27. Lau MS, Mooney PD, White WL, Appleby V, Moreea S, Haythem I, et al.
Prevalence of celiac disease in at-risk and not-at-risk groups in the United Erratum to: ‘Pre-endoscopy point of care test (Simtomax-IgA/IgG-
States: a large multicenter study. Arch Intern Med. 2003;163:286–92. deamidated gliadin peptide) for coeliac disease in iron deficiency
3. Mustalahti K, Catassi C, Reunanen A, Fabiani E, Heier M, McMillan S, anaemia: diagnostic accuracy and a cost saving economic model’. BMC
et al. The prevalence of celiac disease in Europe: results of a centralized, Gastro­enterol. 2016;16:122.
international mass screening project. Ann Med. 2010;42:587–95. 28. Mooney PD, Wong SH, Johnston AJ, Kurien M, Avgerinos A, Sanders DS.
4. Rubio-Tapia A, Ludvigsson JF, Brantner TL, Murray JA, Everhart JE. The Increased detection of celiac disease with measurement of deamidated
prevalence of celiac disease in the United States. Am J Gastroenterol. gliadin peptide antibody before endoscopy. Clin Gastroenterol Hepatol.
2012;107:1538–44. quiz 7, 45 2015;13:1278–84.e1.
5. Hill I, Fasano A, Schwartz R, Counts D, Glock M, Horvath K. The preva- 29. Leffler D, Schuppan D, Pallav K, Najarian R, Goldsmith JD, Hansen J,
lence of celiac disease in at-risk groups of children in the United States. et al. Kinetics of the histological, serological and symptomatic responses
J Pediatr. 2000;136:86–90. to gluten challenge in adults with coeliac disease. Gut. 2013;62:996–1004.
6. Singh P, Arora S, Singh A, Strand TA, Makharia GK. Prevalence of celiac 30. Raju SA, Mooney PD, Aziz I, Kurien M, Sanders DS. Letter: gluten chal-
disease in Asia: a systematic review and meta-analysis. J Gastroenterol lenge in the era of noncoeliac gluten sensitivity – a change in clinical
Hepatol. 2016;31:1095–101. practice? Aliment Pharmacol Ther. 2016;43:656.
7. Mora M, Litwin N, Toca MdC, Azcona MI, Solís Neffa R, Battiston F, 31. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac
et al. Prevalence of celiac disease: multicentric trial among pediatric disease: time for a standardized report scheme for pathologists. Eur J
population from five urban districts in Argentina. Arch Argent De Pedia. Gastroenterol Hepatol. 1999;11:1185–94.
2012;110:490–6. 32. Benkebil F, Combescure C, Anghel SI, Besson Duvanel C, Schäppi MG.
8. Al Hatlani MM. Prevalence of celiac disease among symptom-free children Diagnostic accuracy of a new point-of-care screening assay for celiac
from the Eastern Province of Saudi Arabia. Saudi J Gastroenterol. disease. World J Gastroenterol. 2013;19:5111–7.
2015;21:367–71. 33. Bienvenu F, Besson Duvanel C, Seignovert C, Rouzaire P, Lachaux A,
9. Yap TW, Chan WK, Leow AH, Azmi AN, Loke MF, Vadivelu J, et al. Bienvenu J. Evaluation of a point-of-care test based on deamidated gliadin
Prevalence of serum celiac antibodies in a multiracial Asian population – peptides for celiac disease screening in a large pediatric population.
a first study in the young Asian adult population of Malaysia. PLOS One. Eur J Gastroenterol Hepatol. 2012;24:1418–23.
2015;10:e0121908. 34. Bienvenu F, Anghel SI, Besson Duvanel C, Guillemaud J, Garnier L,
10. West J, Fleming KM, Tata LJ, Card TR, Crooks CJ. Incidence and preva- Renosi F, et al. Early diagnosis of celiac disease in IgA deficient children:
lence of celiac disease and dermatitis herpetiformis in the UK over two contribution of a point-of-care test. BMC Gastroenterol. 2014;14:186.
decades: population-based study. Am J Gastroenterol. 2014;109:757–68. 35. Raivio T, Kaukinen K, Nemes E, Laurila K, Collin P, Kovács JB, et al. Self
11. Rubio–Tapia A, Kyle RA, Kaplan EL, Johnson DR, Page W, Erdtmann F, transglutaminase-based rapid coeliac disease antibody detection by a
et al. Increased prevalence and mortality in undiagnosed celiac disease. lateral flow method. Aliment Pharmacol Ther. 2006;24:147–54.
Gastroenterology. 2009;137:88–93. 36. Korponay-Szabó IR, Szabados K, Pusztai J, Uhrin K, Ludmány E,
12. Makharia GK, Mulder CJ, Goh KL, Ahuja V, Bai JC, Catassi C, et al. Issues Nemes E, et al. Population screening for coeliac disease in primary care
associated with the emergence of coeliac disease in the Asia-Pacific re- by district nurses using a rapid antibody test: diagnostic accuracy and
gion: a working party report of the World Gastroenterology Organization feasibility study. BMJ. 2007;335:1244–7.
and the Asian Pacific Association of Gastroenterology. J Gastroenterol 37. Almazán MV, Ortega E, Moreno Torres R, Tovar M, Romero J,
Hepatol. 2014;29:666–77. López-Casado M, et al. Diagnostic screening for subclinical celiac
13. Talley NJ, Walker MM. Celiac disease and nonceliac gluten or wheat disease using a rapid test in children aged 2–4. Pediatr Res. 2015;78:
sensitivity: the risks and benefits of diagnosis. JAMA Intern Med. 280–5.
2017;177:615–6. 38. Nemec G, Ventura A, Stefano M, Di Leo G, Baldas V, Tommasini A, et al.
14. Leonard MM, Sapone A, Catassi C, Fasano A. Celiac disease and nonce- Looking for celiac disease: diagnostic accuracy of two rapid commercial
liac gluten sensitivity: a review. JAMA. 2017;318:647–56. assays. Am J Gastroenterol. 2006;101:1597–600.

The American Journal of Gastroenterology www.nature.com/ajg

 Point of care Test (igA/igG-DGP) for celiac Disease 9

39. Laadhar L, Kallel-Sellami M, Zitouni M, Mehrezi A, Makni S, Ben Hariz transglutaminase antibodies, a new approach for diagnosis of celiac
M. Is the rapid whole blood test useful for diagnosis and monitoring disease. Indian J Gastroenterol. 2017;36:481–6.
celiac disease in children? Tunis Med. 2011;89:16–7. 49. Kaukinen K, Turjanmaa K, Maki M, Partanen J, Venalainen R, Reunala
40. Baviera LC, Aliaga ED, Ortigosa L, Litwin N, Peña-Quintana L, Méndez T, et al. Intolerance to cereals is not specific for coeliac disease. Scand J
V, et al. Celiac disease screening by immunochromatographic visual Gastroenterol. 2000;35:942–6.

COLON/SMALL BOWEL
assays: results of a multicenter study. J Pediatr Gastroenterol Nutr. 50. Coburn JA, Vande Voort JL, Lahr BD, Van Dyke CT, Kroning CM, Wu
2007;45:546–50. TT, et al. Human leukocyte antigen genetics and clinical features of
41. Singh P, Wadhwa N, Chaturvedi MK, Bhatia V, Saini S, Tandon N, et al. self-treated patients on a gluten-free diet. J Clin Gastroenterol. 2013;
Validation of point-of-care testing for coeliac disease in children in a 47:828–33.
tertiary hospital in north India. Arch Dis Child. 2014;99:1004–8. 51. Aziz I, Lewis NR, Hadjivassiliou M, Winfield SN, Rugg N, Kelsall A, et al.
42. Lewis NR, Scott BB. Systematic review: the use of serology to exclude A UK study assessing the population prevalence of self-reported gluten
or diagnose coeliac disease (a comparison of the endomysial and tissue sensitivity and referral characteristics to secondary care. Eur J Gastro­
transglutaminase antibody tests). Aliment Pharmacol Ther. 2006;24: enterol Hepatol. 2014;26:33–9.
47–54. 52. Kabbani TA,Vanga RR,Leffler DA,Villafuerte-Galvez J,Pallav K,Hansen J,
43. Lewis NR, Scott BB. Meta-analysis: deamidated gliadin peptide antibody et al. Celiac disease or non-celiac gluten sensitivity? An approach to
and tissue transglutaminase antibody compared as screening tests for clinical differential diagnosis. Am J Gastroenterol. 2014;109:741–6.
coeliac disease. Aliment Pharmacol Ther. 2010;31:73–81. 53. Sharma P, Scotland G, Cruickshank M, Tassie E, Fraser C, Burton C,
44. Niveloni S, Sugai E, Cabanne A, Vazquez H, Argonz J, Smecuol E, et al. et al. The clinical effectiveness and cost-effectiveness of point-of-care tests
Antibodies against synthetic deamidated gliadin peptides as predictors of (CoaguChek system, INRatio2 PT/INR monitor and ProTime Micro-
celiac disease: prospective assessment in an adult population with a high coagulation system) for the self-monitoring of the coagulation status of
pretest probability of disease. Clin Chem. 2007;53:2186–92. people receiving long-term vitamin K antagonist therapy, compared with
45. Rashtak S, Ettore MW, Homburger HA, Murray JA. Comparative useful- standard UK practice: systematic review and economic evaluation. Health
ness of deamidated gliadin antibodies in the diagnosis of celiac disease. Technol Assess. 2015;19:1–172.
Clin Gastroenterol Hepatol. 2008;6:426–32. quiz 370 54. Gliddon HD, Peeling RW, Kamb ML, Toskin I, Wi TE, Taylor MM, A sys-
46. Abrams JA, Brar P, Diamond B, Rotterdam H, Green PH. Utility in tematic review and meta-analysis of studies evaluating the performance
clinical practice of immunoglobulin a anti-tissue transglutaminase and operational characteristics of dual point-of-care tests for HIV and
antibody for the diagnosis of celiac disease. Clin Gastroenterol Hepatol. syphilis. Sex Transm Infect. 2017;93:S3–15.
2006;4:726–30. 55. Hoerter NA, Shannahan SE, Suarez J, Lewis SK, Green PHR, Leffler DA,
47. Beltran L, Koenig M, Egner W, Howard M, Butt A, Austin MR, et al. et al. Diagnostic yield of isolated deamidated gliadin peptide antibody
High-titre circulating tissue transglutaminase-2 antibodies predict small elevation for celiac disease. Dig Dis Sci. 2017;62:1272–6.
bowel villous atrophy, but decision cut-off limits must be locally validated. 56. Lau MS, Mooney PD, White WL, Rees MA, Wong SH, Kurien M, et al.
Clin Exp Immunol. 2014;176:190–8. The role of an IgA/IgG-deamidated gliadin peptide point-of-care test in
48. Venugopal G, Mechenro J, Makharia G, Singh A, Pugazhendhi S, predicting persistent villous atrophy in patients with celiac disease on a
Balamurugan R, et al. Sequential testing with different tissue gluten-free diet. Am J Gastroenterol. 2017;112:1859–67.

© 2018 The American College of Gastroenterology The American Journal of Gastroenterology