 Good afternoon. So I'm going to shift gears and speak a bit more granularly about much of the work that I do, which is in the molecular profiling of cancer. Everyone knows about precision medicine, that's why we're here. So in the cancer context, what we're mostly talking about today is looking at genetic changes to establish diagnosis or prognosis, importantly response to therapy and potentially side effects. So in our traditional analyses, we're looking at tumor, our traditional analyses are tumor type and marker specific, we're using them in histochemistry, in situ hybridization and PCR, we're looking at a single or a small number of markers with highly specific limited information. An example would be her two testing here, we're looking for an increase in her two copy number that predicts responsiveness to its Tuzumab and you can do that a few ways, fish, histochemistry or even expression profiling. Another example would be this single mutation test for EGFR mutations where we're looking at mutations in a single gene that relate to a single class of drugs and predict therapy. But what's happened is we've come to look at cancer in a new way and now we're seeing cancer as a series of interrelated pathways and what we're trying to do now is look at many genes and interrogate them for mutation simultaneously to try to understand the biology of the tumor and to treat it in that manner. And so hence we have next generation sequencing, which is the tool that allows us to do this. It's resulted in enormous decreases in per base costs. We can interrogate large regions of DNA, targeted regions. We can look at the whole exome or genome. We can do deep sequencing to detect low level mutations. And the detectable mutation types span the gamut from single nucleotide changes to rearrangements and copy number changes as well as deletions and insertions. So basically, this is really entered routine care in a big way. And there's a couple of ways to do this. This is a report from Foundation Medicine, which is one of the early providers in this area. Since I believe 315 genes and a bunch of introns, so this is one way to do it. They use a capture method for library construction. And then there's, I think, the more common method within academic medical centers and hospitals and amplicon-based method of generating the library, which really looks at tumor hotspots. Typically these are anywhere from, say, 20 to 50 genes. So how are our oncologists using these tools? Well for diagnosis, and this tends to be somewhat more in the diagnostic realm. I think it's even more so perhaps the pathologists, where they use these as ancillary markers to help identify tumor versus benign or tumor subtypes. So we're looking at appropriate targeted therapies, the identification of resistance mutations, but truthfully, much of the broad testing. And I'll distinguish broad testing from simply using NGS as a methodology. At our institution, if I do an EGFR, I'm going to do it by NGS, that's what I do. And that's just a methodologic substitution that allows us to do a bunch of mutations that are known and enhance the workflow. But the big use for broad profiling is actually off-label use in clinical trial selection. And what we're finding is a lot of variants, large numbers of mutations with potential prognostic and therapeutic relevance. We often don't know what, usually don't really know what to do with them. Many genes overlap different cancer types. Germline variants are really key issue that we're just beginning to address. These must be distinguished. So to translate all of this into a useful test requires the ability to first accurately and reproducibly detect variations, which is another whole talk in and of itself, but then meaningfully interpret the results and effectively communicate them. And the types of questions we ask are, is the mutated gene potentially relevant to the patient's management? If so, in what way? Is the particular variant potentially relevant to the patient's management? If so, in what way? The variant analysis can be equally or more complex than the gene analysis at the ground level. And then is it the type of gene that appears to respond to therapy? Is it a type of mutation that appears to respond to therapy? Is it an entire synkinase domain? And sometimes we do a lot of extrapolation. So how do insurers and how do payers view this? Well, this is our, I think the country's largest private payer. So this is a recent coverage decision, July 1st. Molecular profiling using multiplexer next generation. Sequencing technology has proven medically necessary for guiding systemic chemotherapy in patients with metastatic stage 4 non-small cell lung cancer when the following criteria are met. And basically they say testing for EGFR, her two mutations, RET rearrangements and ALC gene rearrangements. That doesn't sound like 315 genes. It doesn't sound like 50. Their comment at the bottom is molecular profiling using multiplex or NGS technology is unproven and not medically necessary for all other indications. So that, and I think that, by and large, I think that many payers perceive what we're doing in that way. So are they right? I think they have an argument to be made. So we did a study at our institution. We did a prospective study looked at, I think, 224 tumors that were profiled. We found, I think almost half of them had some sort of potentially actionable mutation. In the end, about 11%, I think 24 out of the 223 patients received some therapeutic change. Half of those were clinical trials. That was 12. Nine of those were off-label use. And then three of them were on-label use, which, you know, anything on-label you could, you know, pretty much get with other tests. So to the extent that these are patients who have exhausted other therapeutic options, they're prognosis isn't good. We're basically using and pathophysiologic reasoning, educated, informed decision-making, in order to try to treat people who don't have other options. Including putting them on clinical trials shouldn't ensure pay for that. You know, they may consider research, but I think at a major cancer center, people come to a major cancer center to be treated that way. How can we fill these gaps? Really what we need are more data. And I think the problem is that the data are lacking. Again, we have the situation with patients who have extremely bad, extremely poor likelihood of a positive outcome. They've exhausted therapeutic options. We really don't have anything else we can do for them. Can we find a mutation that they're potentially responsive to? Do we have a targeted therapy that is active against a tumor type, a different tumor type with that mutation? Can we try it in another tumor? And I think that's sort of what we're left with there. Can we put them on a clinical trial, which in some ways is, you know, from a medical information standpoint is better. The new type of trial, the new innovation in trials that we're looking at our basket trials, instead of looking at the histology of the tumor, we take tumors and we look at the mutation. And we categorize them by mutation and drug. So NCI has a couple of these types of trials that they're funding, the NCI impact trial. In this trial, they're looking at four treatment regimens. They're looking at three pathways and 20 targeted genes. There's the NCI match trial where they're looking at potentially 3,000 patients. Again, a series of different tumor types looking at specific mutations and specific targeted drugs. This is a different type of study, a registry study. The ASCO is a taper trial that's taking place in private practice. I'm actually on the, we have a molecular, I'm on a molecular tumor board. We have this, we actually have meetings with community physicians who have questions with patients who are enrolled in the study and present the findings to a tumor board to decide whether or not to take action with a specific drug. So what do we need? We need trial information, but we also need tools to help the people providing these data to efficiently and effectively communicate the meaning of the variants. This is an example of one such tool, Alamute. I use it, it provides links to different resources and sort of helps with organization. Believe it or not, Google can be a really good tool. I use it all the time. It's probably better than anything else we have out there. But that, and then this resource, my cancer genome, I think a lot of you are probably familiar with this from Vanderbilt that William Powell I think was the originator and this provides a wonderful summary of the meaning of various mutations in different tumor types. And then we need to learn how to report. And to collate these data, so this is a company and of one they provide a decision support product. It helps list clinical trials in addition to, so it will give you the mutation, help you draft a report which provides a mutation and potential therapeutic options and then potential clinical trials. This is another company that does something similar and they break it down by region. It's always hard to get people on clinical trials. It's people don't necessarily want to go to, people who live in Cleveland say they don't want to go to, you know, can't go to a trial in Los Angeles. So we try to use tools that can inform the oncologists about local trials. And so I think in the end what we're really dealing with is a situation where we have wonderful tools, we have to use them to learn, we don't have any other option. The reimbursement is questionable at best for the bigger panels. The smaller panels are starting to be reimbursed but in the solid, I saw one series of data from one Medicare provider, 75% of the profiling and the 5 to 50 genes, the 81450 code was paid 75% of the time, around half or less for the solid tumor code. So we're starting to get some inroads in more narrow profiling. There's some tests before the FDA with smaller numbers of markers. And I think what we're really finding is that the key markers and the ones you can develop a solid evidence base for each tumor tend to be relatively small in number. The problem is that there can be reasons to look at more markers in an individual case. And it also, from a laboratory perspective, it's very helpful to be able to perform a test, one test for everything unless you have enormous volumes. So, but I guess what I, the take-homes that I would leave from this presentation is really, I mean we have to acknowledge I think that there is a positive data but we also have patients who are, who have exhausted therapeutic options. We have a core group of mutations in many tumors that are known to be actionable. And we have a series, and a growing number of over 1,000 targeted therapies in development for other mutations. Can you go from one tumor to the next with a particular drug? That's just something we're going to have to empirically learn. So I would, I will acknowledge that the data aren't there presently to definitively decide whether or not this testing is going to help somebody, whether in bulk or even in, you know, many of the specific genes and mutations. But on the other hand, we're not going to learn if we don't do it and patients don't have any other options and if we do have a realistic and reasonable chance of helping those patients, I would urge a broader consideration of the clinical utility criteria that many payers, upon which many payers rely to make these decisions. Thank you, Roger. Next we have Julie Johnson presenting from the Knight Network.