Hirschsprung disease (HSCR) — the most common hereditary cause of intestinal obstruction — is in itself an obstructive disease to study. It is clinically variable, shows non-Mendelian inheritance and is associated with several genes, all of which, including RET , EDNRB and SOX10 , are involved in neural-crest development. Mutations in these genes explain the transmission of the long-segment form of HSCR (L-HSCR), but generally not that of the more common, non-syndromic, short-segment form of the disease (S-HSCR). The two forms of HSCR differ in the extent to which agangliosis — caused by the failure of neural-crest-derived ganglionic cells to migrate into the intestinal tract — occurs in the intestine. Aravinda Chakravarti, Stanislas Lyonnet and colleagues now report the genetic dissection of S-HSCR in an approach that could be a model for studying other complex diseases.

The study began with a genome scan of 49 S-HSCR families, including 67 distinct affected sibpairs (ASPs), in which the authors tested both genome-wide polymorphic markers and L-HSCR-associated genes for linkage to S-HSCR. They detected statistically significant allele sharing among ASPs (by identity by descent — IBD) at three loci: 10q11, 19q12 and 3p21. Neither 19q12 nor 3p21 has been previously associated with HSCR. Because RET maps to 10q11, the authors screened the studied families for RET mutations and found 17 of them, but only in 40% of RET-linked families, indicating that non-coding RET mutations might contribute significantly to S-HSCR. Moreover, nearly all the RET mutations mapped to the protein's extracellular domain, in contrast to the gene-wide RET mutations seen in multi-generational HSCR (S-HSCR has a lower familial incidence than L-HSCR). In 27 of the families, one IBD RET allele was shared, which, in 21 of these families, was maternally transmitted — a transmission bias that might explain why HSCR shows a greater than expected inheritance through the maternal line.

Further analyses of the linkage data revealed that segregation at all three loci is sufficient and necessary to explain S-HSCR. Moreover, the expected frequency of individuals heterozygous at each loci — the most common at-risk genotype — closely matches the observed incidence of the disease.

So, how might these loci interact to cause S-HSCR? The authors tested four possible models — additive, multiplicative, mixed multiplicative and epistatic — and found that the multiplicative model, in which the combined effects of all three loci cause the disease, provides the simplest explanation of disease clustering. As S-HSCR usually does not segregate in the absence of RET, the other two loci are probably RET modifiers.

Although several questions remain to be answered — such as the identity of the genes at 19q12 and 3p21 — this comprehensive study sheds important light on the genetic architecture of S-HSCR. It also importantly provides new solutions to the long-standing problem of identifying the contribution of specific genes to a complex disease.