 You know, this is going to be a progress report. If you bothered to read the abstract that was submitted, it was very generic, didn't have much in the way of data, but I'm happy to say that in the month or so since the abstracts were submitted, we have in fact made great progress, and I'll try to describe some of that for you today. First of all, a few facts about bladder cancer. 90% over age 55, half over age 73, four times more likely to occur in men than in women. The lifetime risk for men is 1 in 27, for women 1 in 85. It's the fourth most common cancer in men. And the U.S. spends $2.2 billion a year in healthcare for bladder cancer patients compared with $1.4 billion for prostate cancer. If that isn't a blatant cry for attention and for funding, then I don't know how to do it differently, but it was surprising to me that that is in fact the case. Low-grade tumors are superficial, less likely to invade or metastasize, frequently reappear after resection, but are amenable to therapy and the mortality rate is quite low. In contrast, high-grade, as you would expect, has a propensity to invade and metastasize, the mortality is high when it is invasive, good response to treatment if detected early. In terms of the TNM staging, this picture says it all. The focus of this project is muscle-invasive form of the disease. That is stage two to stage four. Muscle-invasive bladder cancer is 15 to 20 percent of patients with the early stage progressed to muscle-invasive, but 80 percent of patients with muscle-invasive cancer have presented de novo with those late stages. Distant metastases are most common cause of treatment failure. They appear to be present at the time of cystectomy. They occur in 40 to 50 percent within two years without additional therapy. In other words, this disease is not as familiar to many of us as prostate cancer, breast cancer, but it is a major killer and poses considerable problems for us in terms of therapy. Cisplatinum-based multi-agent chemotherapy is the standard of care, both for neo-vagivant therapy prior to cystectomy and for treatment of measurable metastatic disease. Surprisingly, perhaps, there have been no new FDA-approved drugs for muscle-invasive bladder cancer in over two decades. What we see in terms of the survival statistics is suggested by these two plots, one comparing neo-agivant and control and the other metastatic disease treated either with genocytobine cisplatin or with another cisplatin-based regime. The basic message here is that there are a percentage who do pretty well with this disease after muscle invasion, but we don't really know very much about the 20 percent who do and the 80 percent who don't. TCGA's projects obviously aren't set up and designed to answer that particular question, but one can hope that the kinds of molecular information we're developing will be helpful, at least at the level of clues, if not statistics, in arriving at better therapy. Now, here are the leaders of the working group. Seth and I are the co-chairs. I would like to emphasize the role of the coordinators. First of all, Chad Creighton, who's done a superb job of rounding up the data. He's in the terminology the data wrangler, and it also provided much of the material for this presentation. Also, Rehan Abani on the analysis side has done a fine job, as has Margie Sheth, among her 20 different tumor types that she's handling. She has somehow found time to do a remarkably good job with us. Manuscript coordinator is Maggie. Sorry for the spelling. There's no E on Maggie. Morgan from Baylor, who has helped us to put together text. And then, when we saw early on from the first 28 samples as a result of work by J.Galkin co-workers, that the chromatin remodeling was a major theme in bladder cancer as it was in renal cancer and as it's appeared in the other cancer types, we established a special subgroup for chromatin remodeling studies under David, Jonathan, and Peter. Of course, there's a much larger group who have been involved one way or another in this working group. I must say that the same as with other working groups, some have really been working, some have been deeply involved, and others have been less so. Clinical data, then, on the TCGA samples per se. Again, as I said, muscle invasive urethial cancer, allowing mixed histology up to 9%. 126 samples in the data freeze that we are intending for the marker paper. 23 qualified 138 in the pipeline when I say that we're intending for the freeze. This is a little different from what you've seen with respect to other projects because we haven't gotten to our first data freeze. I'll come back to that in just a moment. In terms of gender, male 72%, females 28%, Caucasians 85%, median age 69 with a wide spread, and follow-up highly variable but with a median of 209 days. In terms of the events, 10 progressed. 35 deaths for 126 patients. The progression was out of 31. Staging, shown here, you'll see that the majority were stage 3 and no negative. There is one who's listed as stage 1 and exactly what the status of that particular sample in relation to this study is awaiting pathology review. Here is the incidence of regional node involvement. Most, as I said, are node negative. And here the AJCC stage is pretty much evenly divided among stages 2 to 4. Smoking history is important in bladder cancer as in several of the other cancer types or I guess all the cancer types. Out of the 126 patients, as you'll see, the most common were reformed smokers for more than 15 years, or at least those who claim to be reformed smokers for more than 15 years. So the status of the project. Cruel was a limiting factor, but recently accelerated thanks to hard work and cracking of the whip by Kenna and Seth Lerner, who provided much of the material that I've described to you on the clinical aspects of this project. We had a productive face-to-face meeting in Houston in October that is last month. I'll come back to that at the end of the talk. The data freeze at 126 samples plus normals was decided upon that will probably be in about two weeks. The SNP calls are the last and validations of the last things to come in along with the pathology reevaluation. This is a little different from what you've seen from other sorts of projects, even the thyroid project at least had a month or so after the first data freeze. We decided that in order to move things along quickly that we would do as much as we could toward the marker paper and toward mature analysis as early as possible, even if we knew that we didn't have the final data sets. So the data analysis presented here will be what in the world of clinical trials might be called an interim look. You'll be seeing them on all the data slides with the term preliminary analysis which is meant to signal to you that there is more mature information to come. We are making fast progress on the marker paper and I'll describe that to you later. So on to the data. Here is what you will recognize as the usual significantly mutated gene compendium. I will point out MLL2 which will come back in later part of the discussion. TP53, KDM6A, ARID1A, other familiar genes that you've seen with other tumor types are frequently mutated. Here are the just calls and again I will just leave this up for a moment. I'm not going to try to describe it in detail since it is interim. You'll see again some of our old favorites represented here. Here's an unsupervised clustering of the methylation data. You'll see that there seem to be four categories here. It was given to us as unsupervised and I would have to ask Peter whether it is representative of all of the genes or genes unbiased filtered or not but in any case this is the clustering as it came up. As I say on the lower right hand of these slides there will be a best attempt to give credit where credit is due. An mRNA analysis based on subtypes showing apparently three. This is definitely a supervised analysis by Yuxin Liu and Wei Zhang. And also here is, I'm sorry this is mislabeled, it's not superclustered, it's mRNA clustering supervised in relation to genontology terms on the right and I'm sorry this didn't come out well in the reproduction are the FDRs for the ability of genes in each of these go categories to distinguish between these two apparent clusters of samples. This is from Katie Hoadley and Billy Kim. Here is unsupervised consensus clustering of the microRNA data using this kind of figure before in the other talks and previously seemed to be most likely four different clusters by this analysis from the British Columbia group. Here is a clustering of the RPPA data principally by Rehan Akmani. You'll see at the top which you probably, some of you in the front will be able to make out but I don't want to stress it, are bars which represent the relationship between this clustering and a number of both technical batch effect possibilities to explain clusters and then other sorts of analyses that have been done. This is represented on poster that you probably saw yesterday. Here is the supercluster result, that's a new algorithm that Rehan developed for putting together information from a number of different types of data in this case mutation, microRNA, copy number, DNA methylation, mRNA, RPPA. It shows if you stretch the imagination a bit two to three clusters but not really a dramatically good correspondence among the different types of information in terms of what they're suggesting about the categories of these samples. Here is a vignette on fusion protein. This is FGFR3-TAC3 from Raju's laboratory. Xiaoping Soo has similar data showing this kind of fusion protein. In fact there are a large number of fusion proteins that have shown up that remain to be validated and what I'm showing you here is just an example of a very prominent example of such fusions. Splice variation is a significant feature of these data sets. This is an analysis due to Mike Ryan using his elegant splice seek program package. In this case it's showing the splice variation in CD44 that is exon skip in heavy smokers as opposed to non-smokers. Here are the statistics compiled by that program. What you see in red is the heavy smokers, green is the non-smokers and there is an exon skip event which is prominent in the smokers but not the non-smokers. Again there are many different apparent splice variations that show up some interesting stories in that for future tellings. Viral integration is significant in this case. This is from analysis by Xiaoping Soo. Four samples have integration sites for four different viruses, HPV16 which is of course important for cervical and other cancers. 45 and 56 variants and also BK. This is from the first 85 samples ready for analysis. Other three samples with viral sequences don't have any detected integration sites. That's for HPV6 and CMV. Now you'll note that in a couple of cases DEC1 is the integration site. As an aside we'll point out to you a problem if you are laundering your data sets on genes at any time through Excel. Excel in its infinite wisdom will turn DEC1 into one hyphen DEC thinking it's a date. This is irreversible. You can't retrieve the original form and it's hard to get around if for example you're cutting and pasting into Excel even if you try to format the fields in Excel appropriately. There are about 30 genes we found a while back that have this problem in Excel so they may be lost to your searches forever if you're not cognizant of the fact. Incidentally this program package, Viraseak principally from Xiaoping Soo just came out a few days ago in bioinformatics so you may not have seen it yet. It is useful. There's another set of viral integration studies being done by Raju's group and there will be others I assume. So chromatin remodeling. This is a very interesting story that has developed. I don't know how exactly to give the credit to the working group except to show you the title from Tautoshi's slide which you can still see today represents individuals from a number of institutions in case you can't read it. Here it is in large print. Peter Larratt is the senior partner in that effort. I'm going to show you just a bit of it. It's too complex a story to go into in depth. Here what you see is epigenetic modifiers mutated in more than three samples out of the hundred analyzed to that point. You'll see again if you peruse the labels on the left if you can see them in the front that there are some old favorites are 1A, MLL, 2, MLL, 3 among them. And here are the five that reached significance by statistical criteria. Obviously these are not all that may in fact be significantly mutated if we have larger sample size. I'm going to be focusing on MLL2 since that's a novel story. MLL2 mutation has not previously been reported in bladder cancer specifically. So here is the information on MLL2. It's a complex and very interesting story that involves for example a difference between methylation of the body of the gene and of the promoter region of the gene rather than trying to tell you that story. This is information that is again on the poster that is presented by Toshi. You can look at it, talk with Peter or talk with Toshi about it in detail. And finally among the data slides this is our draft pathway figure put together by Chad Creighton. You'll recognize the format of it. Again I'm not going to try to go through it in detail given the time available but you'll see that it involves the P53RB pathway, the RTK Brass PI3 kinase pathway, histone modifications which are part of the chromatin story and so forth. Well those are the kinds of data that are being rapidly developed even before our data freeze. What I like to do in closing is to say just a few words about how we're approaching writing the marker paper. It's an idiosyncratic approach in some ways and it may as a story be useful to other groups who are approaching that task or perhaps other groups who already had the experience can tell us why we're wrong and the way we're doing it. As I said accrual was slow, accelerated recently. We wanted to get as fast a start as we could toward the marker paper as soon as a reasonable amount of the data, the majority of the data for the freeze were available. So what we did last month at our workshop was to ask for volunteers to write sections of the paper and volunteers to lead in developing the figures for the paper. We don't want these to be cookie cutter papers but on the other hand if you go back through the papers published to date there are certain common features. There's a section of text and an accompanying figure on mutation and copy number. There's a section on expression, et cetera. So in fact four out of five of the figures fit a stereotype as to what kinds of information one wants to include. And similarly I went back through the text of the papers, counted the words that were devoted and the sections that were there for various pieces that canonically appear in the marker papers and then we assigned those or people self-assigned themselves to be responsible for turning those in and in fact within a few days of the time that we put the screws to them to have it ready and this took us up to yesterday we in fact got all the contributions in terms of draft sections of the paper and draft figures for four out of five figures for the paper. So we had word counts and assigned reasonably approximate word counts meant that freed those who were developing these sections from having to write pages and pages supplements to come later and this task seems to this point at least to have gone very smoothly. So now the next stage when we get the actual data freeze will be to update the analyses and then to perform the task integrating all of them and integrating the text so it doesn't look as though it was written by a committee or by disparate individuals. So with that I will stop and be glad to take questions we have a few minutes left for questions. Thanks John, that was a great talk. In other cancers the presence of a virus has occasionally been linked to a specific mutation spectrum. So I wonder in this case have you seen a difference in mutation spectrum in those cases with virus present? It's an interesting question. Is Shalping here? Yeah, so as Gaddy says, that's actually the finding that led us to find viruses in the bladder cancer. Yeah, as I suggested multiple groups have been looking at this and glad to say that maybe for once the results appear to be quite concordant. So John, I actually wanted to ask about a different finding which is the report of CDKN1A mutations. So CDKN1A encoding P21 is one of the sort of growth suppressive genes that's never been seen before to be mutated in cancer. And I wonder if you can comment on the nature of those mutations if they look like tumor suppressor mutations and if they're mutually exclusive with P53 mutations or not. Jaigal, are you here? Is someone from that group who's looked at those data closely? Yeah. Yeah, CDKN1A is the mostly truncating mutation so it's a loss of function mutations but we couldn't find the significant mutually exclusive between CDKN1A and TPP3. Any other questions? Okay, we're ahead of schedule. Sorry, yes. Can you just briefly comment on the frequency of the structural variation in FGF3TAC3? To comment on the frequency. Yeah, it's very frequent, very rare. Right now it's a 3 out of 85. Yeah, other questions? All right, we are a couple of minutes ahead of schedule.