This is the third post of my mini-series on the common susceptibility region for coronary heart disease (CHD) and type 2 diabetes (T2D) on human chromosome 9p21.3. If you have missed the previous parts, they can be found here: I and II.
We have seen previously that the authors of the CHD papers have mapped the risk region for coronary heart disease to a short interval of 58 kb, which contains the three SNPs with the strongest risk association (rs1333040, rs2383207 and rs10116277). One of the papers cites HapMap data arguing that this region is part of a larger 190 kb linkage block. I don’t quite understand why this should be relevant though; both groups have tested lots of SNPs within this 190 kb block, all of them showing much weaker association (if any) to the disease risk. This means that the linkage in this 190 kb block is not that great, at least not in the population used for the studies. I would support the idea that the e 58kb region from coordinates 22,062,301 to 22,120,389 harbors the causative polymorphism. We should keep in mind, however, that the causative variant doesn’t have to be one of the SNPs mentioned above – it can be anything else in that region. Fortunately, at least one of the CHD papers has sequenced the entire region for both high-risk and low-risk alleles, so there is a chance of identifying the real culprit.
All papers on type 2 diabetes have found the strongest association for the SNP rs10811661, which lies just outside (centromeric) of the 58 kb region. Unfortunately, the three consortia distribute their efforts over several risk regions, and the follow-up analysis on T2D linkage to the 9p21 region is not as detailed as that for CHD. The T2D SNP outside of the CHD region, but only by less than 4 kb. Neither the CHD nor the T2D papers report on the association status of the ‘foreign’ SNPs. This information would have been most useful for judging if the underlying cause is a single polymorphism increasing the risk of both diseases, or if there are two different polymorphisms. First, let us have a look at the different options:
- There is a single causative polymorphism that increases the risk of both CHD and T2D
- There are two different polymorphisms in the same gene, one increases CHD risk, the other T2D risk
- There are two different polymorphisms in different genes that just happen to be close to each other.
Did I forget anything? Next, let us have a look at the facts and see which result supports which option. The authors of the different studies probably have better data, but I must rely on what has been published.
- The two types of studies found non-overlapping sets of SNPs. This argues against option 1 and supports options 2 and3.
- The risk-associated SNPs for T2D and CHD are neighbors, separated only by 4 kb. This argues against option 3 and supports options 1 and 2.
- CHD and T2D are ‘similar’ diseases in the sense that they share many common risk factors. This fact supports a connection, i.e. options 1 or 2.
- Study participants were selected in a way to exclude known risk factors such as obesity. Also, CHD patients were not part of the T2D study and vice versa. As a consequence, the diseases will appear uncorrelated in the group of study participants. In an interview (cited according to the NYT article) Kari Stefansson of DeCode is quoted saying that the two diseases are uncorrelated, suggesting that the close proximity of the susceptibility markers might be a mere coincidence. This statement clearly favors option 3. I don’t know if the conclusion is based on data published in the papers, or if deCode has additional data pointing to option 3.
So, what is the most likely explanation? I don’t know. What follows now is pure speculation. I would consider scenario 2, or a modified version of scenario 1 as being most likely. I find it hard to believe that the risk for two similar diseases should be associated with two neighboring genes at such a short distance by mere coincidence. Thus, I’d rather believe that only a single gene is involved. It is possible that – as in scenario 2 – two different polymorphisms change the properties of a gene in two different ways (either qualitative or quantitative), predisposing a subject to one or the other disease. It is also possible – as in a modified scenario 1 – that there is a common polymorphism that predisposes a subject to a certain condition (undiagnosed, hence not a known risk factor) which in turn increases CHD or T2D risk depending on other factors (genetic or environmental).
If we believe the statement that there is no correlation between the CHD and T2D risks, the latter scenario requires a mechanism that prevents patients from getting both diseases. However, if I understand the study setup correctly, even a strong linkage between the two diseases could make them look uncorrelated: the exclusion of diabetes patients from the CHD studies and vice versa could artificially deplete the pool of individuals with a strong inter-disease linkage.
In the next post of this series, I will have a look at the potential genes found in the risk interval. Before I conclude today’s post, let us not forget about scenario 4: Maybe some kind of Intelligent Designer has chosen to put the two risk genes into close proximity, just to keep us scientists busy. Who knows? And I won’t even start talking about the predisposition to taking Avandia…