rs17166496 — FSTL4 FSTL4 rs17166496

Intronic variant in FSTL4 (Follistatin-Like 4) on chromosome 5q31; heterozygous GC carriers show a ~23% reduction in type 1 diabetes risk in the WTCCC genome-wide association study, though the 5q31 causal gene remains disputed due to extensive regional linkage disequilibrium

See your personal result for FSTL4

Upload your DNA data to find out which genotype you carry and what it means for you.

Upload your DNA data

Works with 23andMe, AncestryDNA, and other DNA test exports. Results in under 60 seconds.

FSTL4 rs17166496 — A Signal in the 5q31 Immune Landscape

The chromosome 5q31 region is one of the most gene-dense immunological hotspots in the human genome, packed with interleukin genes (IL4, IL5, IL13) and other immune regulators across hundreds of kilobases of strong linkage disequilibrium. rs17166496 sits within an intron of FSTL4 (Follistatin-Like 4), a secreted glycoprotein that modulates BMP signaling11 BMP signaling
Bone Morphogenetic Protein signaling; BMP ligands regulate cell differentiation, inflammation, and tissue homeostasis and is predominantly expressed in the brain and central nervous system. The variant was identified in the landmark 2007 WTCCC genome-wide association study as a moderate-evidence signal for type 1 diabetes susceptibility — specifically, one where the heterozygous genotype carries a meaningful protective effect.

The Mechanism

FSTL4 is a member of the follistatin-like protein family, which includes FSTL1, FSTL2 (IGFBP7), FSTL3, FSTL4, and FSTL5. These proteins share structural domains — a Kazal-like domain, follistatin modules, and EF-hand calcium-binding domains — that enable them to bind and inhibit BMP ligands and other TGF-β superfamily members. FSTL1, the most-studied family member, has established roles in immune regulation and cardiac protection. FSTL4 and FSTL5 remain among the least-characterized proteins in the family.

rs17166496 is an intronic variant (NM_015082.2:c.727+19462, intron variant on the coding strand) located deep within FSTL4 and does not alter the amino acid sequence. Its functional mechanism — if any — is presumed to be regulatory: the variant may affect intronic splicing enhancer sequences, alter chromatin accessibility, or modify the expression of FSTL4 or a neighboring gene in immune-relevant tissues. The FSTL4 gene has low but detectable expression in the thymus (1.5 nTPM) and bone marrow (2.2 nTPM), tissues central to T-cell education and immune self-tolerance. However, a causal link between rs17166496 and FSTL4 expression in these tissues has not been established.

The chromosome 5q31 locus has been observed across multiple immune-related GWAS studies with extended linkage disequilibrium making it difficult to isolate the true causal gene. The WTCCC paper22 WTCCC paper
Wellcome Trust Case Control Consortium; Genome-wide association study of 14,000 cases. Nature, 2007 explicitly notes that "the identity of the causative gene is disputed owing to extensive regional linkage disequilibrium" at this locus.

The Evidence

The primary evidence for rs17166496 comes from the 2007 WTCCC genome-wide association study33 2007 WTCCC genome-wide association study
Wellcome Trust Case Control Consortium; n=14,000 cases across seven diseases; 3,000 shared controls; Affymetrix 500K array. Nature 447:661-78, 2007, which was the largest GWAS of common diseases at the time and remains one of the most influential studies in human disease genetics. rs17166496 appeared in Table 4 (moderate-evidence associations) for type 1 diabetes (T1D):

  • Heterozygous CG genotype: OR = 0.77 (95% CI 0.68–0.87) — a protective effect
  • Homozygous CC genotype: OR = 1.09 (95% CI 0.92–1.29) — not significantly different from GG
  • Genotypic p-value: 5.20×10⁻⁶ (below GWAS threshold of 5×10⁻⁸)
  • Control allele frequency: G=39.1% in British controls

The protective effect is most pronounced in the heterozygous CG state — a pattern suggesting a dominance effect where a single C allele is protective but two C alleles lose this advantage. This non-additive inheritance pattern is biologically unusual and, alongside the sub-threshold p-value and lack of independent replication at genome-wide significance, warrants caution in interpreting the finding.

The WTCCC results should be considered generating rather than confirming evidence. The 5q31 signal has been noted in T1D linkage studies dating back to the 1990s, suggesting biological relevance, but no subsequent study has definitively identified a specific causal gene or variant in this region for T1D at genome-wide confidence. The evidence level for rs17166496 specifically is therefore rated as emerging.

Practical Actions

For people carrying the GC genotype (heterozygous), the data suggest a modestly reduced T1D risk compared to GG homozygotes, consistent with the observed protective odds ratio. This does not translate to a specific supplement, dietary change, or lifestyle intervention — the mechanism is unknown and there is no evidence that external factors modify the genetic effect at this locus.

For GG homozygotes (the genotype with the reference-level T1D risk in the WTCCC data), awareness of family history and standard autoimmune diabetes screening is appropriate, especially if first-degree relatives have T1D. Autoimmune antibody testing (islet cell antibodies, anti-GAD, anti-IA-2) is the established approach for identifying pre-diabetes in genetically at-risk individuals.

Interactions

The 5q31 region sits adjacent to major cytokine genes IL4, IL5, and IL13 — all Th2 immune regulators — and upstream of the IRF1 (interferon regulatory factor 1) gene region. Extended LD means rs17166496 may be tagging a functional variant in any of several neighboring genes rather than a FSTL4-specific effect. Known T1D risk genes such as PTPN22 (rs2476601) on chromosome 1p13 and ERBB3 (rs2292239) on chromosome 12q13 represent independent loci with much stronger and better-replicated associations. rs6596075 is another chromosome 5 variant that has been noted in the same WTCCC study context at 5q31 in a gene-desert region and is in partial LD with rs17166496.

Genotype Interpretations

What each possible genotype means for this variant.

GG Baseline Risk Normal

Reference genotype with standard type 1 diabetes susceptibility at this locus

You carry two G alleles at rs17166496 in the FSTL4 gene. In the WTCCC genome-wide association study, the GG genotype served as the reference category for type 1 diabetes risk comparisons. GG homozygotes do not carry the protective effect seen in GC heterozygotes (OR 0.77). The G allele is the GRCh38 reference allele and appears in approximately 39–41% of Europeans, making GG the minority homozygote (roughly 16% of European individuals).

CC Major Allele Homozygote Normal

Homozygous major allele with no significant type 1 diabetes risk difference from GG

You carry two C alleles at rs17166496 — the major allele in Europeans, carried by approximately 35% of people of European ancestry. In the WTCCC T1D analysis, CC homozygotes showed an OR of 1.09 (95% CI 0.92–1.29) versus GG — not statistically significant and consistent with no meaningful difference in T1D risk between CC and GG. Notably, CC does not carry the protective effect seen in the GC heterozygous state.

CG Protective Heterozygote Beneficial

Heterozygous genotype associated with reduced type 1 diabetes risk in the WTCCC study

You carry one G allele and one C allele at rs17166496. In the 2007 WTCCC genome-wide association study (n=~2,000 T1D cases, ~3,000 controls), the CG genotype was associated with a 23% reduction in type 1 diabetes risk compared to GG (OR 0.77, 95% CI 0.68–0.87). This is the most common genotype in Europeans, carried by approximately 48–50% of people of European descent.

The protective effect appears specific to the heterozygous state — CC homozygotes did not show the same reduction. The mechanism is not established, and the p-value (5.20×10⁻⁶) is below standard genome-wide significance thresholds. Treat this finding as informative but not definitive.