So in the last lecture we thought about hypermethylation and how single gene biomarkers could be used prognostically, diagnostically and/or to treatment outcome. But now we're going to think about how you can have hypermethylation and measure hypermethylation of sets of genes or panels of genes and how these might be useful. Want to start out by telling you about a particular tumour type called CpG island methylator phenotype, CIMP, as a way of describing the hypermethylation of sets of genes and what it might mean. So these cancers, these have got more than you would expect. So, higher levels of CpG island hypermethylation than you would see in general in cancers. And so this methylated phenotype, has been, has been called CIMP in the literature. So, this frequent occurrence of hypermethylation has been described for several different types of cancers. It was first described for colorectal cancer. Later for glioma, more recently for neuroblastoma and many more. So in each of these cases we know that although in, for each of the tumour types you see hypermethylation. That's very frequent, hypermethalation of CpG islands, that's quite common in them and perhaps more so than in other tumours. The actually sets of CpG's that are hypermethylated differ by tumour type. So colorectal cancer won't necessarily have the same set of CpG islands that a hypermethylated as you see in glioma, for example. So why would we care if they're a particular cancers that have this hypermethylation phenotype, this methylated phenotype or CIMP? Well, as usual with cancer, the reason that we care is because these types of tumours are clinically different. They're clinically distinct and therefore, alter the way that they can be interpreted by the doctors, and therefore the outcome for the patients. So, if we start out with Colorectal cancer. And just think about Colorectal CIMP, because this is where we know them most. It was when it was earliest described for the first time. We know that in general in these patients, they tend to be of older patient age, they're mostly females. They tend to have defective MLH1 function, this mismatch repair gene which I've told you about with regard to Lynch syndrome. The tumours tend to be in the ascending colon. They even have a particular place that they occur in the colon. And importantly, they have good prognostic outcome. So, if you can diagnose a person as having CIMP then you can tell them that they are that you can infer that they are likely to have a good outcome, and this is really important for patients. But also these clinically distinct phenotypes, these clinically distinct features of CIMP suggest that perhaps some particular epigenetic therapy would be useful in these cases. So just like we spoke about single bio markers in terms of being useful for diagnosis and prognosis and for monitoring panels of bio markers are also very useful in this way. So where single genes can be very useful if they're are a perfect bio marker if you like. if you have panels of bio markers, each individual bio marker may not be as strong as say MGMT was for determining whether temozolomide would work, or GSTP1 in the prostate. But each of them can be slightly less efficient or slightly less sensitive, but as a panel you can have overall they'll end up giving you a very efficient and very sensitive, read out of where the situation is occurring. So, they're now becoming more commonly used, and more commonly these panels of biomarkers are being put together so that you can still have useful diagnostic information, for example. So, with these panels of biomarkers, you might be able to identify a tumour type. Say, for example, be able to determine that it was a CIMP tumour, because of the level of hypermethylation. How frequently hypermethylation occurred. So other than just tumour type you may also be able to look tumour sub type. So because of these genome wide studies which I mentioned earlier that have occurred for DNA methylation, we now know it's been fairly recent data where they can lets look genome wide at the where the CPG islands are hypermethylated. And they find it's not just possible to say this is a CIMP tumour or not. But rather, you're able to say this is this particular tumour sub type. And so if you stratify each of the tumours based on their methylation profile, genome wide, you can very, very accurately determine which sub type of a tumour is found. And what's intriguing is that this definition of which sub-type you have is perhaps more efficient when using DNA methylation biomarkers than if you use gene expression. So that's intriguing because you would imagine that the DNA methylation marks would be just mirrored by gene expression, but indeed perhaps not all of these methylation markers are mirrored by gene expression and therefore DNA methylation ends up being a little more powerful. So, what I think is perhaps one of the most interesting uses of having a panel of biomarkers, is that in patients where they don't actually know what their original tumour is, you might be able to work out what their primary tumour was. So, in to my mind, if you think about it naively, you think that probably know what their primary tumour was. But this isn't in fact the case. There are a large number of patients that present each year in any country around the world where they don't necessarily know what the primary tumour was. By the time they've been ill enough to come and present to a doctor they already have metastasis, so clinically it's very important to work out which was the primary tumour and which the metastasis. Because it's the primary tumour that will usually define the type of treatment that will be required. And the metastases will relate to that primary tumour. So say for example, if the primary tumour was in a location where you wouldn't detect a lump. Or you wouldn't necessarily begin to feel ill quickly, it will often have metasicised before you felt sick enough to go to the doctor. So in these cases, it's very useful to be able to work out what that primary tumour was. We also know that these panels of biomarkers just like detecting which particular cancer it is, whether its a particular cancer subtype for diagnosis, can be useful for prognosis. So, while I've mentioned that CIMP has a favourable prognostic outcome. It tends to have a very good outcome. The patients respond well to treatment. High methylation in other cancers for example myelodysplastic syndrome this is found in the blood. It's of the myeloid cells. Or lung cancer. High methylation in both of these cases has a poor prognostic outcome. And this is perhaps what you'd expect based on what I said about how the methylation progresses over time. And so each of these are important measures to make. And finally, as I've mentioned before, by using single biomarkers or indeed panels of biomarkers, you're able to monitor the tumour burden, or measure whether the tumour is declining or perhaps again recurring again over time. And this is because of the cell free DNA or because of tumour cells that can be found in the blood and easily monitored. And again, as I mentioned just before, if you use panels of biomarkers, you are more sensitive to be able to take this more accurately as opposed to a single biomarker. So, while we in general have hypermethylation of CpG islands associated with tumour suppressors, it is the most common type of hypermethylation found in tumour cells. What's been recently discovered is you also have hypermethylation of the regions surrounding those CpG islands. These are called the CpG island shores. So, if you imagine that your CpG island is in a certain block and this is, tends to be unmethylated but in a tumour cell will be methylated. The regions surrounding the CpG island. The 2 kilobases upstream or downstream. It's been recently discovered that although they don't have the same density of CpGs, therefore they're not still considered the island, they are also hypermethylated in cancer. So, this has also been shown in normal cells. So in normal cells, the methylation at the promoter can predict whether or not a gene will be expressed or not. So if, if a promoter is hypermethylated, its unlikely to be expressed. It's not necessarily true that a hypomethylated promoter will be expressed, but if it's hypermethylated, it's highly unlikely to be expressed. And it's known now that this methylation actually does indeed spread to these shore regions. This happens both in normal cells and in cancer. What's interesting again is that if you correlate the hypermethylation at the shores rather than just at the CpG island, this perhaps correlates even better with gene expression than CpG island methylation itself. And so this leads to the possibility that perhaps although I've told you these biomarkers that are at the CpG islands are actually really good, maybe in the future we'll be able to do even better. So the final sort of hypermethylation I want to mention in the context of cancer is hypermethylation of imprint control regions. So one of the really common features of cancer cells is they display loss of imprinting. So in other words, genes that should be displaying monoallelic parent-of-origin-specific expression, in other words, they're imprinted, no longer show this imprint of expression but rather they become either expressed from both parental alleles or silent from both parental alleles. And the reason this is found in cancer is probably because lots of genes that are imprinted are involved in growth - in some way. Either they're growth promoting or growth suppressing. So we don't exclusively find hypermethylation of imprint control regions in cancer. You can also find hypomethylation of imprint control regions, and this of course depends on the particular imprint control region you're looking at and the function of the genes that are found within the controlled region. But in this case I just want to mention the hypermethylation of imprint control regions with one specific example. But keep in mind that either can occur in fact, that we can have hypermethylation in some instances. And hypomethylation in other imprint control regions. So if you think back to the Igf2H19 cluster that we spoke about in week 4, we know here, here is the imprint control region shown as the oval. And it's methylated on the paternal allele, and it's unmethylated on the maternal allele. When it's unmethylated CTCF will binds this insulator element, and it means that the enhancers in this case will act on H19. But Igf2 will be silent for the maternal allele, so we don't see expression. On the paternal allele, because this is methylated, now the enhancers can act on Igf2, because CTCF is not binding to inhibit this, and IGF2 is expressed from the pattern allele. However, with loss of imprinting and what happened is you have hypermethylation of the imprint control region on the maternal allele as well, now, on the maternal allele, you also have expression of Igf2. So now you have a double dose of Igf2 in comparison to what you saw in a normal cell. And Igf2 is growth promoting, and this is associated with Wilms’ tumour. This is this particular childhood kidney tumour that is observed. So, while I'm showing you these for these one specific example, loss of imprinting is observed in a very wide array of tumour types. And in fact, occurs as a very early event, so it's often seen in preneoplastic tissues. So, before you would actually decide that it really was a frank tumour. And so this suggests that maybe these events at the imprint control regions, these hypermethylation, or hypomethylation events might be very early events in tumourogenesis. So, we've now covered hypermethlyation in the context of cancer and in the next lecture we'll think about hypomethylation that's observed so the less methylation that you'd expect which is found genome wide.
