Click the banner above to watch the interview in full!

Jeff Elton: It’s a great pleasure to have George Sledge with me. Thank you very much for being here today. What I learned from the Precision Oncology Alliance meeting is that I should not try to introduce you, it was absolutely unnecessary, because you’re one of the rock stars of ASCO. But, I do want to confirm just confirm one thing, is it true that Taylor Swift is going to have you do a set with her tonight?

George Sledge: You have obviously never heard me sing.

Jeff Elton: It’s circulating around, so I need to understand that. We’re going to go through a few different areas. I think we’ll talk about the Caris approach as one part of what we’re doing. And we’re actually going to start by talking about how clinical genomics data, in different ways, can change how you might plan clinical trials, translational strategies, and some other approaches.

What we’ll try to do is we’ll leave some time at the end for questions. George, just to begin, I know that Caris and ConcertAI have come together and we spent a lot of time thinking about our own partnership and relationship, and part of our enthusiasm was what was so differentiated about the approach that Caris was taking to molecular diagnostic testing. I’m wondering, as a starting point, and you being slightly newer in your chief medical officer executive role coming in here, how would you describe what’s different in terms of the approach relative to other commercially available, next generation sequencing-based approaches?

George Sledge: Sure. If I may, let me start by explaining why I came to Caris because I think it’s a good part of the answer to that question. I’m a lifelong academic. I spent four decades in academia. Throughout my career, I’ve been a data hog. Like many academicians, I loved having larger and larger data sets. Within the last decade, for the first time in the history of oncology, we’ve been able to get data sets that are comprehensive, intense, data rich and attached to clinical data in such a way that we can actually think about using it for the first time to affect the lives of patients. For me, that’s been one of the more exciting things in my life. I came to Caris, in large part, just simply because Caris, I thought, did it better than anyone else.

If you look over the last decade, where for the first time we’ve been able to do next generation sequencing, most of the work that has been done has been done with limited panels. Panels that are in the tens or at most the 100s in terms of the genes one is looking at. By and large, most of these panels have been DNA-based. Caris, several years ago, made the strategic decision, but, I think, also the patient-devoted decision, that more data was better.

Rather than working with a limited gene panel, we would look at the whole exome and the whole transcriptome. In other words, all 23,000 genes that make up every cell in our body and the transcripts coming off of those genes. Having that data is incredibly powerful. No one knows where next year’s drug target will be. If you are stuck with a panel that is looking just at last year’s drug targets, you’re missing something. At Caris, we have, in essence, future proofed things by virtue of just collecting so much data. The ability to use that data, of course, has obvious research purposes, and has obvious purposes in terms of being able to develop signatures in terms of therapeutics for which drugs will work in which patients. These are all things that I think you can only do with large data sets.

Jeff Elton: When you think about it, taking a full exome and transcriptome, was actually only found in smaller scale academic data sets, but without high depth in the clinical portion of the record. What’s the combined value now for both the genomic and the clinicals as we start thinking about almost population scale?

George Sledge: Again, I approached this from the past of having looked at decades worth of research where people would try and come up with a particular signature for, say, “I’m a breast cancer doctor, how does hormonal therapy work in breast cancer?” Number one, what became very obvious many decades ago is that if you look at a single gene, you never got the whole story. Number two, if you looked at small populations, the likelihood that you would have a false positive when you did your data analysis was high.

Jeff Elton: Okay.

George Sledge: For decades we’ve lacked decent validation studies. You have a literature, it’s polluted. Pollution is probably the right term for it. Polluted with studies, done by well-meaning people, but coming up with totally false results. Having large databases on the molecular side that are attached to large clinical databases allows you to get around many of those problems. There’s an old saying that quantity has a quality all of its own. I think that’s absolutely the case when we’re talking about molecular databases and their attached clinical databases.

Jeff Elton: The word polluted is a strong concept, and I can’t ignore that. Let’s come back to that here for a moment. That would almost indicate that, without calling out any one domain or study area, that some of what has been in the published literature that may constitute a body of assumed insight may in fact need to be reconfirmed. Or, we may need to now take advantage of actually just a completely different scale and depth of data and reconsider classes of questions and areas just because of that.

George Sledge: Indeed. Now, some of the failures are, in essence, statistical failures. The P value was 0.05 on a population of 100 people. The next study with another 100 people, it was 0.02. The following study, it was 0.23. You never really knew. Part of that is just simply a mathematical problem. But, I think the other part is a conceptual problem around biology. That’s to say, again, most of the history of our field is looking at single factors, at single genes, single mutations, single transcripts. This fails to capture the complexity of biology. In particular, it fails to collect the complexity of tumor biology, where we’re having to deal with dynamic evolution of the tumor.

Jeff Elton: If I go over to the nature of translational sciences, we were talking a little bit about this before we started the session. If you’re talking about a company that may be taking a novel target, going into an area that may be implicated in multiple tumor types, how would you think about beginning to use some of these data to plan what those first in-human studies may look like? And how might you think about changing the approach that you take to the clinical studies themselves? Because it sounds like the data we can now acquire, even standard of care tools, are actually providing more depth than things that were being used in early phase clinical trials.

George Sledge: I guess part of the answer to this is, if you look at targeted therapies, therapies that form the basis of modern precision oncology, pretty much every drug that falls in this category has, what I would call, some form of gatekeeper biomarker. That is to say that, in breast cancer, you measure HER2. Or lung cancer, you’re looking at a ROS or an ALK mutation. These gatekeeper mutations are what gets you in through the gate with your drug. But it absolutely, positively does not tell you whether or not you’re going to respond to that therapy. If you look at something like Herceptin for HER2 positive breast cancer, if you’ve got amplification of HER2 and you get Herceptin as frontline therapy, your overall response rate is 30%. It’s gotten you in the gate. But, most drugs fail most patients most of the time.

This is the dirty little secret of medical oncology. Most drugs fail most patients most of the time. That gatekeeper mutation identifies, more than anything else, a population that won’t benefit from therapy. But it doesn’t tell you for sure who is going to benefit from therapy. The hope, my hope certainly, and I think the hope of all of us in this field, is that collecting more data is going to get you closer to truth and is going to get you closer to knowing who actually will derive benefit from therapeutics. Everyone in this room wants the best for patients. The best for patients is a drug that works reliably, regularly in patients with minimal toxicity and not treating patients who won’t benefit. Having more data, having more molecular data attached to more clinical data, certainly is the best way to do that.

Jeff Elton: George, as I’m listening to you, I’m thinking about both getting in the gate but also understanding that you’re implying that there’s a lot more that needs to be understood to actually increase potential beneficiaries above that 30% mark. You could say that the process of and almost the translational spirit of really understanding that biological system should continue not just through the success of clinical development phases, but actually over the life of the drug now being available for patients themselves. How do you see that changing now with some of the tools that Caris is providing?

George Sledge: Just to just give some simple examples because these are real life examples. If we look at EGFR in lung cancer, we know that the first generation, first couple of generations of EGFR drugs were inevitably thwarted by very specific mutations. Then you have a wonderful drug like osimertinib that comes along and takes care of that problem. But, what you’re actually seeing there is the evolution of the cancer, the mutational evolution of the cancer. For a more recent example, in breast cancer, take elacestrant. If you look at a newly diagnosed breast cancer patient with ER-positive breast cancer, who’s going to go on an aromatase inhibitor and you measure for mutations in the estrogen receptor gene, the SR1, they are very, very rare.

They’re 1% or less of patients. You now put a patient on an aromatase inhibitor for four or five years, and then you look at them when they relapse, that number goes up to maybe a 30 or 40% rate of ESR1 mutations. What do you do now? You get an FDA approval for a drug that works for ESR1 mutations. That’s elacestrant. It was approved about three or four months ago. I think we’re going to see that more and more. Typically, in terms of the life cycle of developing drugs, we also probably should be thinking in terms of patent life because the pharma companies I’ve spoken to over the years tend to have lots of wonderful scientists who are consistently working on a particular disease problem.

Part of the life cycle development of a new drug probably should be what’s going to be the resistance to that drug and how can I thwart it? To do that fairly early in the process, you need to be thinking about doing real-time molecular measurements of the ongoing evolution of the cancer. In the past, for a good part of what we’ve been doing in the molecular diagnostics revolution, we’ve done it off of solid tissue. Increasingly, we’re going to be able to do it off of liquid biopsies, which will probably be a whole lot easier for patients and a whole lot easier in terms of measuring the dynamic evolution of the cancer.

Jeff Elton: I want to come back to this and then I want to spend a little bit of time on the whole area of liquid tumor biopsies. You said something that I think could actually have some reasonably profound implications for how people even think about what their strategies are. Here I’m thinking about a biopharma, because I think what you were calling the gateway mutations probably takes a disproportionate share of the focus and emphasis of most organizations. That’s where a lot of the primary research came through.

But what you’re laying out here is, “if I take a longitudinal view and if I take an exposure based view, I’m going to actually see a range of different transformations and new mutations of interest that may, in fact, be almost as significant as some of the gateway mutations you just identified.” Laying that framework out is not always how people start framing out what they can do and the different strategies they can take and even necessarily what their pipelines look like. That actually in itself, I think, I’d love to follow up with you on because I think there’s some ways that we could begin to publish out some views and information on that material.

George Sledge: Indeed. By the way, this isn’t a criticism of the companies.

Jeff Elton: And that was a real comment, by the way.

George Sledge: This is more a reality of what’s happened. That’s to say, throughout most of the history of drug development, we just simply couldn’t measure this stuff. We could not measure dynamic evolution of human cancers. We lacked the tools. They were too expensive. There were too few places that did it. There was no way to do it in real-time in patients. This is a change that’s occurred, in particular, over the last, I would say, five years or so. It actually allows us, I think, for the first time, to think in terms of the overall lifespan of a tumor and of the patient who has that tumor.

Jeff Elton: You oversee, and I knew about, the area of Caris’ work in signatures. I should let you explain what that term actually means a little bit. I was not aware until I saw your presentation that you have a portfolio that now is under your leadership at Caris. Maybe you could give a little definition and also explain how the data and the information that you now have, combined with almost new methodological quantification and quantitative approaches, are now enabling different kinds of decision support, et cetera. Maybe we could develop this a little bit.

George Sledge: Sure. Let’s start with signatures. I won’t say it’s the opposite of the compliment to the gatekeeper mutation approach. That’s because tumor biology is typically complex rather than simple. By the way, there are some drugs where it’s simple. If you have a BCRA mutation, the biology is pretty simple. The resistance mutations are actually in the same ATP pocket that the original drug tackles. There are cases in cancer where the biology is pretty simple, where, in essence, they’re the stupid cancers. But, the problem is most cancers are smart. Most cancers are complex, the genomics are complex. There’s usually not a single reason for a cancer to become resistant to a drug. There tends to be multiple reasons. Because of that, you need to be thinking in terms of signatures. That’s to say, not a single gene. Not just a gatekeeper gene, but rather a whole bunch of different genes.

How do you discover those? There’s been a number of different processes that have been used. Again, I’m a breast cancer doctor. Probably the most commonly used signature on the planet today is beyond the Oncoype DX signature. 21 gene signature, developed with cutting edge 2002 technology where basically what then genomic health, now exact sciences, did was look at, basically, the entire breast cancer literature and pulled out of that literature everything that had been looked at as a biomarker for breast cancer. They started with 300 different biomarkers. By the way, this gets back to the false positive problem in biomarkers. They then whittled this down to 16 cancer related genes and then weighted them, put them together. That particular signature has resulted in roughly an 80% reduction in the use of adjuvant chemotherapy in breast cancer over the last two decades. Think about that.

Think about the expense saved. About the toxicities saved. Women who don’t have to throw up and lose their hair because of a molecular diagnostic. In particular, it’s because of a relatively more complex signature. The difference between then and now is that then, because you couldn’t look at 23,000 genes simultaneously, you had to start with the list that came from the literature. You don’t have to do that anymore. Now you get some tumor, either solid or liquid, and you measure all of those mutations. You then have very smart mathematicians.

At Caris, we have a wonderful artificial intelligence group that spends an enormous amount of time putting together complex signatures. We’re presenting one of those at this meeting, if I could share that with you. This is one looking at who gets a brain metastasis. For a very long time, one of the tragedies of cancer patients is that we’ll do a great job of treating most of the body, but most of our drugs don’t get into the central nervous system. Therefore, patients will develop a brain metastasis and indeed go on to die of that brain metastasis. By and large, we don’t know who those patients are. By and large, we’ve had no particularly good way of monitoring them. We can’t scan people every three months, in their brain, because it would be cost prohibitive and would drive the patients and the doctors nuts after a while.

But what we think, based on our early studies with work done by Jim Abraham in our group, is that we now have the molecular tools. This, by the way, was developed using 220,000 patients. That allows us to very finely slice the population of solid tumors, not a particular tumor, but in that case, this is multi-cancer. We can identify populations that are at exceptionally low risk for ever developing a brain metastasis and populations that are at quite high risk, maybe three quarters of the patients will develop a brain metastasis in a year and a half. How might I use this? In the clinic, if I have a really high risk population, maybe that’s a population I screen more often. But, if I’m in pharma and I am developing drugs that have CNS penetration, maybe I want something that will allow me to predict, ahead of time, who’s at really high risk for getting a brain metastases in the next couple of years. Then I can now use that population to test the drug’s ability to prevent brain metastasis.

We, for the first time, now have drugs like tucatinib, like T-DM1, that we actually know get into the brain and induce remissions in the brain. Wouldn’t it be wonderful if we can be thinking now in terms of specifically developing CNS penetrant drugs and then applying them to high risk populations and seeing whether or not we might be able to benefit them. Again, we think that the signature we’ve developed might be a first step in that direction. By the way, this is a poster that Jim Abraham has here at ASCO this year. We’re very excited about that. But, we’re in the process of developing a bunch of different signatures. Some of them are specific to particular drugs. Some of them are specific to particular conditions like IO therapy across cancers, for instance. We think just simply by having very, very large numbers of patients who’ve been treated where we have appropriate clinical follow-up data, we should be able to come up with something exciting for many of these scenarios.

Jeff Elton: There are definitely some companies that would be interested in that particular topic. Let me ask you. Your comments about signatures, it’s almost got its own parallel because it sounds like, even if I have an existing therapeutic entity, the ability to think through and use signature-like methodologies could add incremental value to different exposed subgroups and subpopulations and aid in the management. Do you see the notion of signatures almost becoming an integral part of the life cycle of certain oncology therapeutics? Or how do you see that maybe evolving a little bit?

George Sledge: I certainly hope so. Again, in the past, going back to the gatekeeper mutation analogy. Because we didn’t have follow up data, we couldn’t do this because we lack the molecular data and the systems to do this. At Caris, our business model, by and large, is that doctors send us tissues from patients who have advanced disease. Last year, we got 100,000 samples in. We’re on track, I think to do around 140,000 this year. Having that bulk data across large numbers of different cancers gives you lots more options to look at.

On the evolution of cancers, I would add that it also gives you more options to see whether or not your drug works in cancers other than what it was first approved for. Because what frequently occurs out in the community is that a patient has a particular mutation. It’s not one for which the drug was originally approved for. But the doctor says, “What have I got to lose? Let’s try this therapy.” Then we’re actually able to determine, based on follow-up clinical data, if there is evidence of benefit. We’ve actually been looking at that in terms of: can you, in essence, create tissue agnostic indications by using clinical data for broad populations?

Jeff Elton: The FDA actually has some, and some of the more recent guidance documents indicated that models that could become the basis of selection of a cohort may in fact be allowed for certain clinical trial designs. You can imagine these methodologies actually may be a much more robust way, particularly if you’re actually doing it, again, to your point, at a population scale, in terms of where that sits.

George Sledge: To give an example, because this is just a thought experiment that we did. There’s an FDA tissue agnostic indication for pembrolizumab in IO therapy for tumor mutation burden-high patients or DNA mismatch repair. We’ve looked at virtually the same indication for nivolumab and have compared the time on treatment curves for pembro versus nivolumab in the setting of TNB high or DNA mismatch repair. Frankly, the curves are totally overlapping. You could think about: if you have a second or third drug in a class and the first drug has a tissue agnostic indication, could you, using real world scenarios, actually per the FDA guidance, think about using this data to get a broader indication for the drug and therefore help more patients?

Jeff Elton: That should have some people’s attention. Let me take you back to the liquid tumor biopsy. Maybe here, too, Caris is taking a differentiated approach. But first, maybe just a short description on why a differentiated approach here? Also, as you overlay this ability to have liquid tumor data and maybe a neoadjuvant in later phases, what are you seeing in the implications here of at least early views? I’m going to carefully frame it that way.

George Sledge: A liquid biopsy is, in essence, using next generation sequencing of either DNA or RNA. Liquid biopsies have the potential to be used across the entire spectrum of cancer. They have the potential to be used in an early diagnostic setting all the way to late management of patients with metastatic disease. At every step along the way, you can imagine the use of liquid biopsies. Patients don’t like having needles stuck in their liver and lungs repetitively, and doctors don’t like doing it because every time you do it, there’s a potential for hazard.

That hazard doesn’t exist, by and large, sticking a needle into someone’s vein. The potential to monitor is much easier at a patient level, and therefore potentially much more utilitarian in terms of doing repetitive measurements to look at tumor evolution over time. I think that’s the promise of liquid biopsies. Liquid biopsies have their own issues. Some of those are issues are ones that we’re trying to address through our upcoming Caris Assure assay.—CUT—27:54 Similar to our tissue assay, we are doing whole exome and whole transcriptome off of the plasma. But at the same time, —28:09 we also are looking at the buffy coat, the white blood cells. And off of the buffy coat we’ll be looking at the host genome. We’ll also be looking at what’s called chip subtraction. Clonal hematopoiesis, that is to say, mutations that are basically occurring in your bone marrow that are pumping out white blood cells and which increase with age, are really common.

We already know, based on published data – not our data, but other folks’ data – that, for instance, if you look at prostate cancer, a significant proportion of older men have clonal hematopoiesis that actually are mutations that would lead you to give a patient a PARP inhibitor for prostate cancer. In other words, treating a patient’s circulating white blood cells for something that you think might be going on in their metastatic prostate cancer tumor and is not.

If you do a clonal hematopoiesis subtraction, which we’re doing as part of this assay, this allows you to avoid treating patients who will not benefit from the drug. We think that’s going to be a significant addition. This is currently at MoIDX awaiting regulatory approval. We hope that we’ll be rolling it out in the not-too-distant future. If you’re trying to track the evolution of a human cancer, I think this sort of approach, —CUT—29:57  whole exome and whole transcriptome, —CUT—30:01 hugely increasing the amount of data you can get, would be great. But, also from a safety standpoint, doing the subtraction of clonal hematopoiesis will increase the safety for patients with cancer.

Jeff Elton: As you’re talking and also thinking about this longitudinal view, and also understanding what responses looks like and patterns of resistance look like, do you see integrating this early on in a variety of new drug development programs?

George Sledge: Sure. Starting at the very beginning with newly diagnosed patients. One of the great white whales in all of oncology has been cancer prevention. It’s really, really hard to do cancer prevention trials. They’re expensive. They’re time-consuming. A large amount of the expense and time-costs is that if you start with the general population, not many of them actually get cancer. You have to treat a lot of people to benefit a few patients. If you could come up with molecular diagnostics that would allow you to identify those patients regularly, then we’d be able to tighten up the population and make those trials cheaper and quicker. At the time of diagnosis, a woman comes in and has a mammogram, someone comes in and has colonoscopy. If on the day of diagnosis if you could already tell what mutations are present in their cancer, maybe now you’re talking about thinking in terms of molecularly based neoadjuvant therapy starting on the day of diagnosis. The opportunities here, I think, are incredible.

Jeff Elton: Actually, what I’m hearing in that one is historically standard of care tools and standard of care practices probably wouldn’t really support operating that way. But actually, as this now gets more integrated into it, and if we actually have any reason to suspect any form of risk, we could use a variety of other tools. We’ve talked about digital H&Es and other things. There may be ways of using current standards of care to then accelerate where that testing may bring benefits and where those interventions are needed. But, we have some work to do on that.

George Sledge: This is early days for many of these things. Whenever a former academic promises you something, you should ask for several reality checks.

Jeff Elton: Was that a promise?

George Sledge: The excitement in this field, if you were to walk over across the street and into the session, the excitement is palpable. Every session you go into has a new phase three trial for a novel agent that’s a game changer. A large amount of that excitement comes from the fact that this really is approaching real precision medicine, that we are at a patient level actually going to be able to say, “Mrs. Jones, you have lung cancer type 3a and type 3a benefits from this drug. Plan B is that, once we notice the evolution of your cancer to have the mutation, we’ve got another drug for you.” That has been the hope and promise in this field for so long. I think the combination of molecular diagnostics with novel therapeutics actually is getting us close to that.

Jeff Elton: One of the reasons why our two organizations came together was to start creating data sets of a scale and scope that actually further support that. And even being able to get down to narrower populations, but with a statistically significant numbers of patients still in them and with a depth that actually allows us now to be actionable.

George Sledge: We want quantity of data. We also want quality of data. One of the things that makes us particularly excited about working ConcertAI is that it significantly improves the quality of the data that we work with from a clinical standpoint. I think those sorts of combinations, high level molecular data, high level clinical data, get you a lot closer to where you need to be.

Jeff Elton: I do want to offer, if anybody has any questions, we’re happy to take some.

Speaker 4: George, just connecting a few of the topics you cover and that you’re a breast cancer physician, I want to get a bit more specific on drug conjugate and ADCs. You take an ortho. Now antibodies might bring you to the right cells, spread in healthy tissue, so they’re very effective. One side looks like this is a major, major advancement, but on the other hand, you may actually create a veil to pierce through, because then you need to really understand resistance mechanisms. What would be your advice to those who are already treating patients with these drugs and those that are developing?

George Sledge: I thought about this a lot. In breast cancer, we have drugs like trastuzumab, direcsatan, we have isatuximab, govitecan, thoroughly unpronounceable drugs. They’re fascinating in part because they have specific molecular targets, TROP2 or HER2. You’re using the antibody for that. You’re attaching, in essence, a cellular poison to that. What’s the drug resistance based on? We have two decades of research in HER2 about mechanisms of resistance to HER2 targeted therapies. Are those mechanisms still relevant in an era of antibody drug conjugates?

Or is the resistance now basically chemotherapy resistance to the payload? We don’t know, is the honest answer. How are we going to learn that? Again, I would think that having large numbers of patients who’ve been treated with one of these agents and then looking at what happens to the tumor may be the best way of looking at that in allowing us to tease out, “If I give isatuximab, govitecan, does resistance occur because the tumor is no longer expressing TROP2? Or does resistant occur because it’s TOPO-tecan resistance?” And we know a lot about tecan resistance. I think the only way, outside of cell lines in a laboratory which have their own limitations, that we’re going to learn about this is to look at fairly large populations. I think it’s a great question and it’s one I’ve thought a lot about.

Jeff Elton: We may have a follow-up conversation for you on that topic.

Speaker 5: You mentioned tumor evolution a lot. How long is the sample relevant to the patient? Because the diagnostic molecular profile is not going to be the same after two rounds of heavy duty therapies. I know we were talking about the biopsies, but putting that aside, is there a way to avoid needing a relevant sample from a biopsy?

George Sledge: Yeah, it’s a really great question. I don’t know that I can give you a final answer, but I can give you an initial answer. If you look at the correlation between a tissue and a liquid biopsy, because we’ve done this as part of the development of Caris Assure, what you find is that there’s a very high correlation between what you see at a mutational level in the first 90 days or so. If you have the solid tissue biopsy and the liquid biopsy within three months, there’s a really tight correlation between the mutations you see. But, you get more and more drift the further and further out you get. Again, this is exactly what you’re saying there. There’s intervening therapies. Those intervening therapies are affecting the makeup of the tumor.

They’re probably reducing sensitive clones. They’re leaving resistant clones. The mutational makeup changes, and changes pretty quickly. I think what this tells us is that if we’re serious about this, we probably need to start measuring reasonably early and may ultimately need to measure reasonably often. I want to be careful about saying this. This is not George Sledge’s new standard of care, that you need to get Caris Assure every three months for the rest of your life. I think this is going to be very tumor-by-tumor and we’re going to actually need to accumulate the data to demonstrate that this is actually beneficial to patients. One of my mantras is that finding bad news early doesn’t turn it into good news. It’s only useful to do something to the bad news if you can actually alter a therapy and benefit the patients. Again, this is going to be very tumor-by-tumor, very case-by-case.

Jeff Elton: To further support at least the merit of the debate you were having, even some of the ASCO leadership have been advancing that as a goal, so it that may actually become more of a requirement in the near future, really changing outcomes.

George Sledge: The ASCO guidelines are clearly starting to move in that direction.

Jeff Elton: I want to thank everybody for taking the time to be here. I’m super glad that we’ve got this recorded. I really appreciate your perspectives and we really appreciate where the partnership can go to. I know you’ve already identified some research projects that I feel like I owe you some follow-up on. We will get that going as well.

George Sledge: I look forward to it.

Jeff Elton: Thanks very much.

George Sledge: Thanks very much.