Celiac disease genes beyond HLA: what large genetic studies found

Short answer

A 2011 study of Europeans counted 39 known regions outside HLA, and together they explain 13.7% of celiac disease genetic variance1, 2. A 2020 paper says the HLA-DQ gene testing available in clinic does not include non-HLA risk31. Coeliac UK's page for clinicians says no firm conclusions have been made about the importance of the other genes studied, and adds "nor do we routinely test for them"3.

Published by Gluten Verdict · updated · 9 sources, each linked below

Reviewed by Mark Peter, Registered dietitian (RD), on

Person at a library carrel with stacked papers, a highlighter and an orange mug

Key points

  • Celiac disease almost always occurs with DQ2 or DQ8, but about 30 percent of people carry one of them and only about 3 percent of carriers develop celiac disease, per NIDDK7,6.
  • A 2011 study counted 40 known risk loci, HLA included; the 39 outside HLA explained 13.7% of genetic variance there, and HLA about 40% more1,2.
  • Most of the new regions in the 2010 study contain immune function genes12, and many variants sit in non-coding regions, which a 2020 review says suggests an effect on gene expression21.
  • Coeliac UK's page for clinicians says no firm conclusions have been made about the importance of the other genes studied, and adds "nor do we routinely test for them"3; the 23andMe celiac report FDA reviewed in 2017 reads one HLA variant28.
  • For non-celiac wheat sensitivity, a 2020 review says very little is known about the genetics23.

What a genome-wide study is

A genome-wide association study compares how often many genetic variants turn up in people with a condition and in people without it19. Coeliac UK describes one it funded: over 300,000 variants in 778 people with coeliac disease and 1,422 without19.

In the 2011 dense-genotyping study, 29 of 54 fine-mapped signals appeared localized to specific single genes, some of them to gene regulatory elements16.

Two large studies: 2010 and 2011

This section covers two large studies: a 2010 genome-wide association study and a 2011 dense-genotyping study of loci already reported for immune-mediated diseases. The table sets out what each counted.

The 2010 genome-wide study and the 2011 dense-genotyping study side by side
2010 (Dubois and colleagues)2011 (Trynka and colleagues)
People studied4,533 people with celiac disease and 10,750 controls in the first stage912,041 people with celiac disease and 12,228 controls15
Follow-upVariants genotyped in a further 4,918 cases and 5,684 controls10Dense genotyping of 183 non-HLA loci already reported for immune-mediated diseases15
New regions13 new regions reached genome-wide significance1113 new risk loci, bringing the known total to 40 counting HLA1
What it found in themMost contain immune function genes12Multiple independent association signals at over a third of the loci in the study17

Separately, Coeliac UK's account of the study it funded names its first finding: the interleukin-2 and interleukin-21 genes, which it describes as important in determining how strong an immune response is made20.

What the non-HLA regions have in common

In the 2010 study, most of the 13 new regions contain immune function genes, and four of the genes it names (ETS1, RUNX3, THEMIS and TNFRSF14) play key roles in thymic T cell selection12.

A 2020 review says the non-HLA variants are mainly single nucleotide polymorphisms in non-coding regions, often within enhancers, suggesting an effect on gene expression rather than on protein function21. The 2010 paper reports that over half the associated variants were correlated with expression changes in nearby genes13.

The same review notes that several other immune-mediated diseases share susceptibility loci with celiac disease22. A region on a celiac list may therefore also turn up for other conditions.

How much of the inherited risk each part accounts for

Three published estimates give a sense of the proportions. They come from different designs, so they are not one number measured three times.

Heritability and variance estimates from three studies
StudyDesignWhat it estimated
Kuja-Halkola and colleagues, 2016 (Gut)107,912 Swedish twins; celiac disease identified from biopsy reports25Heritability of celiac disease 75% (95% CI 55% to 96%); non-HLA heritability 68% (95% CI 40% to 96%)26
Trynka and colleagues, 2011Case-control: 12,041 cases and 12,228 controls15In Europeans, the 39 non-HLA loci explain 13.7% of celiac genetic variance; HLA accounts for a further ~40%2
Dubois and colleagues, 2010Case-control: 4,533 cases and 10,750 controls in the first stage9The HLA variants plus 26 further loci explain about 20% of total celiac variance, which would be 40% of genetic variance if heritability is 0.514

The twin study's authors write that HLA seems to have only moderate impact on heritability estimates27. The 2011 authors expect a long tail of effects of weaker significance to explain substantial additional heritability18, and the 2010 authors say a long tail of low effect size common variants, along with rare variants, may contribute to the remaining heritability14.

Do clinical tests cover these variants?

Coeliac UK's guidance for healthcare professionals says that of the many other genes looked at as candidates for the extra risk beyond HLA, no firm conclusions have been made about their importance, adding "nor do we routinely test for them"3. It also says genetic testing in the diagnosis pathway is not routinely recommended4, and that HLA typing is used in specific clinical at-risk scenarios, primarily to exclude the condition5. That is a UK charity's account.

One consumer test that FDA reviewed is narrower still. FDA's 2017 review of the 23andMe celiac report says it is based on a qualitative test for a single nucleotide polymorphism, rs2187668, which detects the HLA-DQ2.5 haplotype28. The review adds that the report does not describe a person's overall risk for developing celiac disease29.

A 2020 pilot of a 42-variant risk score

A 2020 paper built a 42-variant genetic risk score from the European case-control study, using HLA-DQ imputation and published genome-wide association studies, then tested it in a pilot of 128 children with suspected coeliac disease32. Its authors write that scores of this kind can discriminate coeliac disease but had not been validated in clinic30, and that HLA-DQ gene testing, which is available in clinic, does not include non-HLA attributed risk31.

In UK Biobank the score did better than HLA-DQ stratification: an area under the curve of 0.88 against 0.8233. The authors conclude it could facilitate non-invasive diagnosis but needs detailed assessment in the context of other diagnostic tests and against current diagnostic algorithms34.

NIDDK says researchers are studying other genes that may increase the chance of developing celiac disease in people who have DQ2 or DQ88. That is research.

Non-celiac gluten sensitivity: what genetic evidence exists

A 2020 review of genes beyond HLA in gluten-related disorders calls the condition non-celiac wheat sensitivity (NCWS). It says that, in contrast with celiac disease, very little is known about its genetics23.

On HLA, the review reports that DQ2 and DQ8 are present in about 40–50% of NCWS patients, a percentage slightly above the prevalence of these haplotypes in the general population, which it takes to suggest that DQ2/DQ8 has minor relevance, if any, in the development of NCWS24. Compare that with celiac disease, where NIDDK says the disease almost always occurs in people with DQ2 or DQ87.

Against that statement, a list of "gluten sensitivity genes" needs a named study behind it23.

Questions to put to a list of "celiac genes"

A list that calls itself "celiac genes" or "gluten sensitivity genes" can be held against what the studies above report:

  1. Does it include HLA-DQ2 or DQ8? NIDDK says celiac disease almost always occurs in people with one of them7, but about 30 percent of people have DQ2 or DQ8 and only about 3 percent of those develop celiac disease6.
  2. Does it cite a genome-wide study with cases and controls, such as the 2010 and 2011 papers9,15? If it names none, there is no study behind it for you to check.
  3. Does it say how much each variant changes risk? In the 2011 study, all 39 non-HLA loci together explain 13.7% of genetic variance2; a list implying one gene decides the matter does not describe that.
  4. Does it claim a clinical use? Coeliac UK's page says no firm conclusions have been made about the other genes and adds "nor do we routinely test for them"3, and the 2020 risk score was a pilot whose authors say it needs detailed assessment34.
  5. Does it treat gluten sensitivity as inherited? For NCWS, a review says very little is known about the genetics23.

Questions people ask

Is there one celiac gene?

Not by itself. NIDDK says celiac disease almost always occurs in people with DQ2 or DQ87, but only about 3 percent of people with DQ2 or DQ8 develop it6, and the 2011 study counted 39 further regions outside HLA1.

How many genes are linked to celiac disease?

The 2011 study counted 40 known risk loci including HLA1. A locus is a region: in that study 29 of 54 fine-mapped signals appeared localized to specific single genes16.

Can a test of these genes diagnose celiac disease?

Coeliac UK says genetic testing in the diagnosis pathway is not routinely recommended4, and its page on the other genes adds "nor do we routinely test for them"3. A 2020 paper says the HLA-DQ testing available in clinic does not include non-HLA risk31. A 2020 pilot of a 42-variant score said it needs detailed assessment against current diagnostic algorithms34.

Are the same genes behind non-celiac gluten sensitivity?

A 2020 review says very little is known about the genetics of non-celiac wheat sensitivity23, and that DQ2 and DQ8 are present in about 40–50% of those patients24.

More in Celiac genetics and family risk

Sources