What the hell is CpG island methylator phenotype (CIMP)? - (Oct/28/2005 )
I was wondering what the hell is CpG island methylator phenotype (CIMP) as proposed by Issa group. Recently they wrote an editoral in Gastroenterology (Issa JP, Shen L, Toyota M.) celebrating somebody else (Samowitz et al Evaluation of a large, population-based sample supports a CpG island methylator phenotype in colon cancer. Gastroenterology. 2005 Sep;129(3):837-45.) finally confirmed their theory.
If I understand correctly, the so called cimp refers to the concordant methylation of a group of genes in cancer and the methylation only occurs in cancer. What is the point in this theory. If you analyze any tumor, you will be able to find out a group of genes concordantly methylated in the tumor and not in normal tissue. In my opinion, cimp is just a fancy term invented to describe something so obvious. Should I coin a term "ear phenotype" to describe the phenomenon of the concordant appearance of two ears on our head?
Just my opinion, but it could be refering to certain forms of cancer in which an oncogene is not mutated but instead a different methylation pattern is the cause of the cancer? I've read some interesting things on topics like this and so called epigenetic inheritence, where you inherit a phenotype that is not the result of an altered genetic sequence but rather due to the identicle coding region containing methylations upstream. Perhaps they are referring to something like this?
My interpretation of CIMP and it is still quite controversial from the conference I was at where Issa presented this, is that CIMP can be used to type a cancer and my well lead towards diagnosing the predisposition of cancer.
So if you take a handful of genes known to be involved in cancer, tumor suppressors, mismatch repair genes and the like. You can identify the type of cancer my looking at the methylation pattern of the CpG islands of these genes. So for example colorectal cancer could be typed as CIMP high meaning that in all the genes selected was looked at the CpG islands are hypermethylated while in another cancer such as prostate cancer the same genes exhibit a CIMP low phenotype meaning that the same CpG islands are hypomethylated.
Having said this, if you take another set of genes and look at colorectal cancer, it could very well be CIMP low while prostate cancer could well be CIMP high.
So the more genes you type for CIMP, the more narrow you can call the cancer type. Sure, this at the moment can only type cancers that have already metastised, a longitudinal study needs to be conducted and is being conducted to see if you can pick up CIMP before the cancer becomes apparent.
very challenging because there is evidence that something other than DNA methylation could be the initial trigger for tumoreigenesis.
hope this has clarified it or maybe muddled the water a bit more. I have met Shen and Toyota and Issa at a conference very very bright people!.
Hi Nick and Captain,
Thanks for the explanation.
Yes I agree that each tumor type has a specific group of frequently methylated genes which we can call methylation signature for a particular type of cancer.
The term cimp still makes no sense to me. For example, they define colon cancer with methylation of p16, MLH etc as cimp+, otherwise cimp-. But for those cimp- tumors, they may have concordant methylation of another group of genes they have not mapped yet.
I believe they are bright people otherwise they won't have invented such term. Last April I attend the AACR annual meeting. There was a minisympoisam session on epigenetics and the first four or five speakers were either from Issa lab or Jones lab, none of their talks impressed me, instead I thought they should have been put to the poster session.
You have a very good point there. You already have the colon cancer, is there a point of defining them as cimp+ or cimp-? as you have cancer already? I think yes there is a point, as the etiology of the cancer could be different in colon cancers that are cimp+ and cimp- for the genes you select. Granted these cimp+ and cimp- cancers could very well be concordant for methylation in another subset of genes, but there is still discordance in the first set of genes.......and the question of why still needs to be answered.
I think CIMP will help in someway as to determine the pathway to cancer tumorigenesis and ultimatly be able to diagnose possible predisposition to a certain cancer type.....because diagnosing for CIMP early and before cancer appears will be very powerful in the prevention of cancer, but this is "blue sky stuff" at the moment!!
I would have loved to have been at the AACR annual meeting, I attended the AACR special conference in Hawaii last year, Issa himself presented as well as Jones and Laird, it was a very high powered meeting in terms of epigenetics, I would have liked to have gone to colorado next year but I have yet to get my results accepted for publication!! grrr
The DNA methylator phenotype refers to a state where DNA methyltranferase activity is abnormal, thus a large(?) number of genes is abnormally methylated. I guess at the moment at least it more specifically refers to hyper-methylation of the "usual suspect" cancer genes (cell cycle, DNA repair etc.)
Thus the methylator phenotype is a term to potentially discriminate between tumours with only a few "sporadic" methylated genes and tumours with systematic hypermethylation of "many" CpG island-carrying genes. Of course you are right to wonder that the more genes you scan the higher the probability to find more methylated ones.
The methylator phenotype follows the RER phenotype example, that was introduced some years ago to describe tumours with DNA repair errors (usually detected in the form of microsatellite instability)at multiple loci.
I guess the term was not presented as a great new piece of philosophy but just to help us talk the same language. On the other hand scientists like introducing terminology :-) but this is not too bad. It is part of the politics :-)
According to Issa group, the definition of cimp+ is:
1. concordant methylation of a group of genes
2. the methylation is cancer specific
If a tumor is defined as cimp+ without referring to what genes are in the group, then cimp+ simply doesn't make any sense because the tumor could be cimp+ for another group of genes.
So my point is only when cimp is associated with a specific group of genes, does it make sense.