HLA Genotype as a Risk Factor for Celiac Disease Among Relatives

In This Article

Abstract and Introduction

Abstract

Background: Family screening has been advocated as a means to reduce the major underdiagnosis of coeliac disease. However, the precise risk of the disease in relatives and the impact of patient- and relative-related individual factors remain obscure.

Aims: To investigate the individual risk of coeliac disease among patients' relatives.

Methods: Altogether 2943 relatives of 624 index patients were assessed for the presence of previous coeliac disease diagnosis, or were screened for the disease. Coeliac disease-associated human leucocyte antigen (HLA) genotype was determined from all participants. The association between individual factors and new screening positivity was assessed by logistic regression.

Results: There were 229 previously diagnosed non-index relatives with coeliac disease and 2714 non-affected (2067 first-degree, 647 more distant) relatives. Of these 2714 relatives, 129 (4.8%) were screening-positive (first-degree 5.1%, second-degree 3.6%, more distant 3.5%). The combined prevalence of the previously diagnosed and now detected cases in relatives was 12.2% (6.3% clinically detected, 5.9% screen-detected). In univariate analysis, age <18 years at diagnosis (odds ratio 1.60, 95% CI 1.04–2.45) in index, and age 41–60 years (1.73, 1.10–2.73), being a sibling (1.65, 1.06–2.59) and having the high-risk genotype (3.22, 2.01–5.15 DQ2.5/2.5 or DQ2.5/2.2 vs other risk alleles) in relatives were associated with screening positivity. Only high-risk HLA remained significant (2.94, 1.80–4.78) in multivariable analysis.

Conclusions: Unrecognised coeliac disease was common among at-risk relatives even in a country with an active case-finding policy, and also in relatives more distant than first-degree. The presence of a high-risk genotype was the most important predictor for screening positivity. ClinicalTrials.gov identifier NCT03136731.

Introduction

Coeliac disease is a gluten-driven chronic gastrointestinal condition affecting individuals with a predisposing human leucocyte antigen HLA-DQ2 and/or HLA-DQ8 haplogenotype.^[1] Estimated prevalence of the disease is up to 1%-3% in general population, but currently, most of the affected patients remain unrecognised.^[2–5] This substantial underdiagnosis could be improved by active testing of either specific at-risk groups or even the whole population with serum coeliac autoantibodies. At present, most authorities do not recommend untargeted screening mainly because of inconsistent data on the prognosis of unrecognised coeliac disease at the population level.^[6–9] However, particularly, the first-degree (FDR) and, sometimes, also second-degree (SDR) relatives of patients are often considered to have a sufficiently high disease risk to justify screening.^[9–13]

Selecting an optimal screening strategy is complicated by a wide variation in the reported family risk^[14–16] possibly due to different poorly defined patient- and relative-related individual factors, such as age at screening, gender, HLA haplogenotype and degree of consanquinity.^[1,16–23] Limited data on these factors make optimal timing of screening, testing of other than FDRs, and the benefits of genetic risk stratification debatable.^{[13,15,16,23]} The heterogenous and often small study cohorts and different diagnostic outcomes in earlier studies further hamper the interpretation of the results and emphasise the need for additional evidence.^[14] Besides optimised implementation of the screenings, better understanding of the individual risk factors could provide novel insights into pathogenesis. In fact, precise risk stratification in coeliac disease is becoming increasingly important as we may be entering the era of primary preventions.^[24,25]

Here we aimed to study the impact of various index patient- and relative-related factors on the risk of coeliac disease. This was established by serological and genetic testing of a large and well-defined cohort of relatives of previously diagnosed coeliac disease patients.

Table 1. Clinical characteristics and HLA distribution in 2714 at-risk relatives with positive ^a or negative screening outcome
	Positive screening, n = 129		Negative screening, n = 2585		P value
	n	%	n	%	P value
Age at screening, median (Q₁, Q₃), y	40 (20, 53)		36 (17, 55)		0.556
Age <18 y at screening	29	22.7	667	26.0	0.404
Women	72	55.8	1435	55.5	0.946
Member of multiple-case family^b	35	27.1	631	24.4	0.483
Degree of consanguinity with index
First-degree relatives	106	82.2	1961	75.9	0.260^c
Sibling	51	48.1	732	37.3	0.073^d
Offspring	33	31.1	783	39.9
Parent	22	20.8	446	22.7
Second-degree relatives^e	19	14.7	515	19.9
More distant relatives	4	3.1	109	4.2
HLA risk group^f					<0.001
High^g	27	26.5	156	7.0
Intermediate^h	75	73.5	1394	62.3
Lowⁱ	0	0	686	30.7

^aPositive endomysium and transglutaminase antibodies and presence of human leucocyte antigen (HLA) DQ2 and/or DQ8.
^bAt least two first- or second-degree relatives previously diagnosed with coeliac disease.
^cFirst-degree vs second-degree vs more distant.
^dAmong first-degree relatives.
^eGrandparent, grandchild, aunt, uncle, niece, nephew, half-sibling.
^fData missing from 376 screened relatives.
^gDQ2.5 homozygotes and DQ2.5/2.2.
^hDQ2.5 heterozygotes or DQ2.2 and/or DQ8 positive.
ⁱDQ2 and DQ8 negative.

Table 2. Logistic regression analysis of different index- and relative-related characteristics for positive screening outcome ^a in 2714 at-risk relatives
	Univariate	Multivariable
	Univariate	Model 1^b	Model 2^c	Model 3^d
	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Index characteristics
Age <18 y at diagnosis	1.60 (1.04–2.45)		1.34 (0.80–2.23)	1.41 (0.80–2.51)
Women	0.73 (0.49–1.07)
High vs intermediate risk HLA	1.37 (0.89–2.12)
Dermatitis herpetiformis	0.91 (0.55–1.52)
Autoimmune co-morbidity	1.39 (0.86–2.23)
TVA vs PVA/SVA	1.02 (0.68–1.62)
Relative characteristics
Age <18 y at screening	0.84 (0.55–1.28)
Women	1.01 (0.71–1.45)
High^e vs intermediate^f risk HLA	3.22 (2.01–5.15)			2.94 (1.80–4.78)
Consanguinity with index
Offspring	1	1	1	1
Sibling	1.65 (1.06–2.59)	1.67 (1.00–2.79)	1.51 (0.88–2.58)	1.30 (0.71–2.35)
Parent	1.17 (0.67–2.03)	1.58 (0.82–3.05)	1.29 (0.61–2.72)	1.05 (0.45–2.44)
Second-degree relative	0.88 (0.49–1.57)	0.86 (0.48–1.55)	0.84 (0.46–1.52)	0.90 (0.46–1.76)
More distant relative	0.87 (0.30–2.51)	0.91 (0.31–2.62)	0.84 (0.29–1.27)	1.24 (0.41–3.76)
Multiple-case family	1.15 (0.77–1.72)
Age at screening, years
0–20	1	1	1	1
21–40	1.06 (0.65–1.74)	0.97 (0.58–1.61)	1.01 (0.60–1.69)	1.06 (0.59–1.91)
41–60	1.73 (1.10–2.73)	1.21 (0.71–2.07)	1.30 (0.75–2.26)	1.62 (0.88–2.97)
61-	0.70 (0.37–1.33)	0.48 (0.23–1.01)	0.57 (0.26–1.27)	0.51 (0.19–1.34)

Note: Characteristics notified in multivariable analysis were ^bconsanguinity with the index and relative's age at screening; ^cmodel 1 and age of index <18 y at diagnosis; ^dmodel 2 and high vs intermediate risk HLA of relatives.
Abbreviations: OR, odds ratio; PVA, partial villous atrophy; SVA, subtotal villous atrophy; TVA, total villous atrophy.
^aPositive endomysium and transglutaminase antibodies and presence of human leucocyte antigen (HLA) DQ2 and/or DQ8.
^eDQ2.5 homozygotes and DQ2.5/2.2.
^fDQ2.5 heterozygotes and DQ2.2 and/or DQ8 positive.