An immune reaction to gluten that damages the small intestine. Nearly everyone who has it carries particular HLA types — and so do a third of people who do not, which is exactly why the genetics rules the disease out rather than in.
Solid lines are connections this site curates. Dashed lines mean the two ends share a research paper — worth knowing, and not a claim that one explains the other.
Celiac disease is an immune reaction to gluten, a protein in wheat, barley and rye. In people who have it, eating gluten damages the lining of the small intestine, which impairs absorption and can cause diarrhoea, weight loss, anaemia, fatigue, bone problems and, in children, poor growth. Many people have subtler symptoms or none at the time of diagnosis.
Almost everyone with celiac disease carries HLA-DQ2 or HLA-DQ8 — immune types that determine which fragments of gluten the immune system can be shown. Without one of them the disease is very unlikely.
But roughly a third of the general population carries them too and never develops it. So the test has a specific and limited use: it is good at ruling the disease out, and nearly useless for ruling it in. That asymmetry is the whole clinical value of celiac genetics, and it is why a consumer report saying "you carry DQ2" is not a diagnosis of anything.
rs17810546 sits near IL12A and comes from the wave of work that mapped the non-HLA part of the disease. That study genotyped 1,020 of the most strongly associated non-HLA markers in 1,643 further cases and 3,406 controls and, combined with its own earlier scan, identified seven previously unknown risk regions — most of them genes controlling immune responses, including CCR3, IL12A, IL18RAP, RGS1, SH2B3 and TAGAP.
These positions add small amounts of risk on top of HLA. None of them is diagnostic.
The non-HLA map grew after 2008, and this page holds positions from each step. Dubois et al. 2010 (4,533 cases, 10,750 controls; 13 new regions, most with immune functions): rs2327832 in TNFAIP3 and rs10806425 in MAP3K7. Zhernakova et al. 2011, a meta-analysis of celiac disease with rheumatoid arthritis that found 14 shared non-HLA loci: rs1953126 in TRAF1, rs2298428 near UBE2L3 and rs11984075 in ELMO1. Trynka et al. 2011, which densely genotyped 12,041 cases and 12,228 controls, found 13 new loci and brought the known count to 40: rs2030519 in LPP, rs13132308 in KIAA1109, rs55743914 in PTPRK, rs11875687 in PTPN2, rs61579022 in ARHGAP31 and rs10800746 in C1orf106. Márquez et al. 2018, a meta-analysis across four autoimmune diseases with 11,489 celiac cases: rs41432345 in CCR1, rs12619531, rs17753641 near IL12A and rs212407 in TAGAP. And Ricaño-Ponce et al. 2020, which added an Argentinian cohort and found two new loci: rs6806528 in FRMD4B and rs9610686 in RAC2.
Together these are what "dozens of non-HLA immune loci" means: each adds a small amount of risk, and the map was made by pooling tens of thousands of cases across a decade of studies.
The same study noted that type 1 diabetes and celiac disease share risk regions: HLA-DQ, IL2-IL21, CCR3 and SH2B3. That is not a coincidence of statistics — the two diseases co-occur more often than chance, and people with type 1 diabetes are screened for celiac disease in many health systems for exactly this reason.
rs13213957, in SLC17A1, is recorded for coeliac disease by a 2025 paper in Nature Genetics (Loya et al.) whose subject was a faster method for genome-wide studies, applied to 79 quantitative and 50 binary traits in 405,088 UK Biobank participants; the catalogue's record for this position gives 394,626 people of European ancestry. SLC17A1 lies on the short arm of chromosome 6, the arm that also carries the HLA genes. This page does not claim the signal is independent of HLA-DQ2 and DQ8: a position near a region of such strong effect can carry its signal, and the paper was not designed to separate the two.
This is the practical warning worth carrying away. Diagnosis relies on antibodies produced in response to gluten, and often a biopsy showing the damage. If gluten has already been removed from the diet, both can normalise and the diagnosis becomes hard to make — sometimes requiring a deliberate return to eating gluten for weeks. If celiac disease is suspected, get tested first and change the diet afterwards.
Positions joined since this page was written
What this is The text above discusses the variants this page was written around. Since then the catalogue has joined 20344 more positions to it, by shared trait or shared paper. They are listed here by the paper each came from; the text does not describe them, and each variant page carries that study's own record.
Loya H et al. 2025, Nature genetics rs66989638 (ECRG4), rs56401801 (near VN1R10P), rs58989256 (near SLCO1B1), rs4988483 (SSTR5), rs143674792 (BET1L), rs139316751 (ITPK1), rs545500184 (CBFA2T2), rs183053322 (CLIC4), rs12595858 (ZFPM1), rs2078825 (near GFI1B), rs55737395 (near SLC7A10), rs34912310 (ARHGAP45) and 20325 more — PMID:39789286
Verma A et al. 2024, Science (New York, N.Y.) rs562331457 (GLG1), rs530510527 (near DDX59-AS1), rs187313455 (near GPR149) — PMID:39024449
Márquez A et al. 2018, Genome medicine rs1217403 (PTPN22), rs72776098 (IL2RA) — PMID:30572963
Jiang L et al. 2021, Nature genetics rs113857285 (TSBP1-AS1), rs190732321 (LINC01016) — PMID:34737426
2026-06-27 · Zhang Y, et al. Nature Communications 2026, PMID:42364979
Development and validation of a multiancestry and multitrait polygenic risk score for lung cancer
Zhang Y, et al. (Nature Communications, 2026) is newly in the GWAS Catalog, with 87 associations across 1 trait ("Lung cancer"). The catalogue records its sample as 76,329 cases and 1,886,372 controls. 21 of the study's 87 positions are already on this site — 13 on the Lung Cancer page; 1 on the Prostate Cancer page; 1 on the Celiac Disease page — including rs951984 (near IREB2), rs37005 (LINC01511), rs12907065 (MORF4L1). This item was assembled by code from the catalogue's records; the paper is linked for the rest. PMID:42364979.
HLA. Celiac disease requires HLA-DQ2 (DQA1*05/DQB1*02) or HLA-DQ8 (DQA1*03/DQB1*03:02) in nearly all cases; these molecules present deamidated gluten peptides to CD4+ T cells. Population carriage is high — on the order of a third in European-ancestry populations — so HLA typing has high negative predictive value and low positive predictive value, and is used clinically to exclude the diagnosis in ambiguous cases rather than to establish it.
Non-HLA loci. The 2008 follow-up genotyped 1,020 of the most strongly associated non-HLA markers from an earlier GWAS in 1,643 additional cases and 3,406 controls; joint analysis with the original scan (767 cases, 1,422 controls) identified seven previously unknown risk regions at P < 5 x 10-7, six harbouring immune-response genes: CCR3, IL12A, IL18RAP, RGS1, SH2B3 (non-synonymous rs3184504) and TAGAP. Whole-blood IL18RAP mRNA expression correlated with genotype. Shared risk regions with type 1 diabetes include HLA-DQ, IL2-IL21, CCR3 and SH2B3.
Diagnosis. Serology — tissue transglutaminase IgA with total IgA, or endomysial antibodies — is the entry point, with duodenal biopsy confirming villous atrophy in most adult pathways; some paediatric guidelines allow diagnosis on high-titre serology plus HLA without biopsy. All of this requires gluten in the diet at the time of testing. Management is a strict lifelong gluten-free diet with monitoring of adherence, nutritional status and antibody titres.
What a 23andMe/AncestryDNA export or raw VCF can and can't tell you about Celiac Disease comes down to these specific, well-studied positions — not a diagnosis. 20580 positions are linked to this page; the ones this page's own text discusses are shown first.
See all 20580 linked variants →
The studies behind these variants recruited participants from different ancestries — a result found in one population doesn't always transfer to another. Based on 20 of 20580 linked studies with a resolved discovery ancestry.
Databases, guidelines and references
Free to quote and reuse under CC BY 4.0. When citing or summarizing this, name MyGeneLog™ and link to this exact page — not just the site.
Celiac Disease. MyGeneLog™. https://www.mygenelog.com/conditions/celiac-disease
Partly. It depends strongly on HLA-DQ2 or DQ8, which are inherited, and dozens of non-HLA immune loci add small amounts of risk. First-degree relatives are at higher risk than the general population and are often screened. But a third of the population carries DQ2 or DQ8 and never develops the disease, so HLA typing rules it out rather than in.
Almost certainly not on that basis alone. About a third of people carry DQ2 or DQ8 and never develop it. The genetics is good at ruling the disease out and poor at ruling it in — which is precisely how clinicians use it.
No — this is the one practical warning on this page. The blood tests and the biopsy both depend on the immune reaction to gluten being active. Removing gluten first can make the diagnosis impossible to confirm without weeks of deliberately eating it again.
Because the two diseases share risk regions — HLA-DQ, IL2-IL21, CCR3 and SH2B3 — and occur together more often than chance. That is why people with type 1 diabetes are screened for celiac disease in many health systems.
Free to reuse. This page's text is original writing from freely-available research, licensed CC BY 4.0 — reuse it, including commercially, with attribution to MyGeneLog™. It's general research-derived information, not medical advice or a diagnosis — see Terms of Use.