Two new association studies have revived hopes that research into unusual monogenic forms of diabetes can provide pointers to the genetic basis of susceptibility to the common multifactorial form of the disease.

Maturity-onset diabetes of the young (MODY) is an autosomal dominant form of the disease that usually develops before an individual reaches 25 years. The hepatocyte nuclear factor-4α gene ( HNF4α ) that underlies one form of MODY encodes a β-cell transcription factor that is involved in insulin secretion. The hope has been that variation in this gene might also help to explain susceptibility to type 2 diabetes (T2D) — the much more common late-onset form.

Despite linkage evidence for a susceptibility locus in the HNF4α region, previous attempts to identify HNF4α variants associated with T2D, which focussed on the coding sequence of this gene, had failed. Now, however, Latisha Love-Gregory and Kaisa Silander, with their respective colleagues, have independently gathered convincing evidence for a link between T2D and variation in the region of an alternative promoter for HNF4α (P2) that drives transcription of what is probably the gene's dominant splice variant in pancreatic β-cells.

Love-Gregory et al. used linkage disequilibrium (LD) to map a 78-kb candidate region that included HNF4α and P2. They identified haplotype tagging SNPs (htSNPS) that would represent the most common haplotypes across this region in Ashkenazi Jews. The authors then compared frequencies of the nine htSNPs in individually genotyped cases and controls of Ashkenazi-Jewish descent. The pay-off for this well-designed candidate-gene-association study was the identification of a htSNP in the P2 region that was associated with T2D.

By contrast, Silander et al. were building on previous mapping work on non-insulin-dependent diabetes mellitus (NIDDM) by the Finland–United States investigation of NIDDM Genetics (FUSION) study on affected sibling-pair families from Finland, which indicated that there might be a T2D-susceptibility locus in the 20q13.12–20q13.13 region. Their approach was to compare the frequencies of 291 SNPs in this region between pooled DNA from FUSION cases and controls. A different strategy in a different population but an equally successful result: another SNP in the P2 region associated with T2D. More intensive follow-up SNP genotyping around HNF4α identified another T2D-associated SNP in the same region, in near perfect linkage disequilibrium with the first.

Commendably, the two groups collaborated to follow up each other's work, and found that the SNPs that they originally identified were in near perfect linkage disequilibrium with each other. Overall, they identified four SNPs in the P2 region that are associated with T2D in both populations and, in doing so, immeasurably strengthened the overall findings.

So, for diabetes at least, those who study rare Mendelian versions of complex diseases in the hope that it will tell us something about the multifactorial disorder seem to be on the right track. However, these results also beg the question: how many candidate-susceptibility genes have been wrongly discarded because only coding SNPs have been examined in association or linkage studies?