 Well, good afternoon everyone. I'm Eric Green, Director of the National Human Genome Research Institute, and I want to welcome all of you to the one and only annual lectureship we host at NHGRI each year, and that's the Jeffrey Trent Lecturing Cancer Research. I'm going to just briefly tell you about the history of this lectureship, and then I'm going to turn this over to Dan Kaster who introduced today's speaker. The intramural program of NHGRI was founded in 1993, when Francis Collins arrived on the scene to be the director of the institute. But upon arriving here, what Francis did was to create an intramural research program at the institute which previously did not exist. And he recruited Jeff Trent, a colleague of his from University of Michigan, to come here and to be the first scientific director, in other words the director of the intramural program at NHGRI. In turn, Jeff recruited a number of people, including myself, to the intramural program, and the rest is sort of history. Jeff was responsible literally from building the program from essentially a non-existent program, recruiting a number of individuals and ably leading the program for the better part of nine years, and really taking it from just complete sort of conception of an idea of having an intramural program dedicated to genomics to actually having been a thriving intramural program that I think has greatly influenced this entire campus through the expertise it's been able to provide in genomics and its application to study human disease and eventually clinical medicine. So when Jeff departed in 2002, I was appointed the scientific director at that time, and one of the first things I did was to start a lecture series that would honor Jeff and to, as much as anything else, just thank him for everything he had done. And so we created the Jeffrey Trent Lectureship, and we focused it on cancer research, and since that time there has been just a remarkable set of lecturers who have come and given that lecture, and it's been just wonderful to see a number of people just sort of come through and give talks really at the inner failure phase of cancer and genomics, something that Jeff was and continues to be passionate about and a real expert in, and it's sort of his legacy in terms of providing leadership to our intramural program very much is reflected by the type of people we're consistently able to attract to give this annual event. So with that as a historic background, I'm going to introduce you to the third scientific director of NHGRI, and that's Dan Kastner, who will introduce today's speaker. Thank you for coming. Well, thanks very much, Eric, and it's my enormous honor and pleasure this afternoon to introduce to you Dr. Catherine Janeway. Dr. Janeway is currently a senior attending physician at Boston Children's Hospital in the Dana-Farber Cancer Institute. Dr. Janeway did her undergraduate work at Barnard in New York, but then went on to Boston where she has been ever since, and she did her medical school training at Harvard Medical School and also got a master's in medical science there at Harvard Medical School as well. She did a pediatrics residency at Boston Children's Hospital in the Dana-Farber and went on to become the chief resident in pediatrics and then did her fellowship in hematology oncology at Boston Children's and Dana-Farber and she has begun to join the staff there at Boston Children's and Dana-Farber and she has become an absolute leader in the application of genomics to pediatric hematology oncology and during the course of the 10 years that she has been on the staff at Boston Children's and Dana-Farber she has become the program director for solid tumors at those institutions. She has become a real leader in Ewing's sarcoma and osteosarcoma. She is the chair, co-chair of a number of different national consortia and collaborative studies looking at ways of improving the treatment of those diseases and she has really been someone who has taken the lead in terms of applying genomics to these very important problems in pediatrics. So, without further ado, I will give you Dr. Jane Way, the title of whose talk will be Bringing Genomics to the Pediatric Oncology Clinic, Diagnosis, Treatment Selection and Rational Clinical Trial Design. Catherine, take it away. Well, thanks very much for the introduction and for the invitation. I'm delighted to be here to speak with you today. So, just to give you a sense of what I'll discuss today, I'll first start by giving you a brief background just to sort of orient you to my perspective on precision cancer medicine in the pediatric oncology space. And then I'll try to sort of concisely wrap up what we've learned from the clinical sequencing studies that have been conducted to date in pediatric oncology and how they inform precision trials. And then I'd like to give you a sense of where we are in pediatric oncology in terms of precision cancer trials. And I'll touch on a couple of different approaches to that. One is to discuss basket trials and relapse disease using the example of the pediatric match. And then I'll touch on two alternative precision trial designs. And then finally, I'll end the talk with a brief discussion of what's going on in the data aggregation or big data analytics space in pediatric oncology. So, I think I don't need to just really convince this audience of the importance or significance of precision cancer medicine. But I pulled a couple of recent examples just to kind of bring it to the front of your mind. And the first on the left here is a very recent publication in the Journal of Medicine demonstrating that in women who have germline BRCA mutations who have newly diagnosed metastatic breast cancer who are randomized to receive either Elaprib, which is a targeted therapy in this disease, or standard therapy, there's a significant improvement in progression-free survival. And Elaprib is a drug that has been studied in the sort of relapse and refractory setting for quite a long period of time in the BRCA deficient cancers. And it's really, I gave this example primarily because it's nice to see this work moving into the upfront treatment setting. And then the second is a very well-known example of really unbelievably dramatic and frequent responses to PD1 blockade in mismatch repair deficient cancers leading to the first FDA approval of a drug for a genomic biomarker. And it's really gratifying to see the sort of beginnings of examples of these types of precision cancer medicine successes in pediatric oncology. This is a slide taken from the ASCO 2017 annual meeting from a presentation of a Phase 1 trial of Lerotrectinib which is a TREC inhibitor in TREC fusion positive malignancies. Turns out that TREC fusion positive malignancies span the age range. And what was incredibly significant for those of us in the pediatric oncology community was to see that the Phase 1 trials in adults and pediatric patients were conducted simultaneously and presented together in a very nice plenary presentation. And this is just a case example of a dramatic response and if you look at the timeline this is the patient at baseline and after four doses of the drug. So it's an incredibly effective targeted therapy and it's really nice to see that our pediatric patients did not need to wait to access this clinical trial. So you know that just hopefully brings to the forefront of your mind why we're here talking about bringing genomics to the clinic and now I'll throw a little bit of cold water on that. So this infographic is beautifully simple and very logical. Patient A has mutation A, gets drug A and has this dramatic response. But I think what we've learned from bringing genomics to the clinic is that the devil really is in the details. It's not that simple. And just to begin there's an incredibly important issue in this field that adds to the complexity is somatic variant interpretation. And so when you draw this nice infographic there are a couple of assumptions that are made, right? One is that mutation A is a well known or well characterized mutation that you know what it's doing to the protein that you know what that protein is doing in that particular cancer type that's sort of undergirded by a huge amount of basic science research that this line between mutation A and drug A is a strong and forceful line meaning there's a ton of evidence that supports a link between this mutation and activity of drug A. But in fact, we really don't we work in a space where when we're bringing genomics to the clinic we often get mutations where we're uncertain about the type of mutation the line drawing the line between this mutation and a drug that's sometimes difficult. And finally, you have to consider the drug itself. Is it really a targeted therapy, right? There are many drugs that are thought to be or that are called targeted therapies that may in fact have off-target activity. And so we really in this field of precision medicine have only just now begun to approach the sort of difficulty of characterizing the somatic variants and what they mean for the patients both in the space. So this I'm showing you now a publication in genome medicine that came from the ClinGen somatic variant working group which basically tries to break this down. And so this is the sort of characterization labels that have been used in the germline space for variants and really they've been posed using these characterizations in the somatic space. And if you have a pathogenic mutation meaning it's likely to alter protein function that you think about the implications in the diagnostic, prognostic and predictive meaning prognostic sorry predictive space. But what's really important in this schema in several other schemas that recommend make recommendations about how to do somatic variant interpretation is the evidence piece, okay? So what do you mean by evidence? So there's evidence about the variant and then there's also evidence about that line connecting the variant to the drug in that example mutation A to mutation B. And that evidence can be preclinical evidence, it can be a case report or it can be a randomized clinical trial. And the reality is that in the pediatric oncology space and also in other rare cancers we're very often operating in a realm without sufficient evidence. I didn't really talk much about germline variant interpretation, I just want to say there are more better established criteria and terminology for variant classification, but germline variant interpretation still requires time and appropriately trained personnel. And so the whole both somatic and germline variant interpretation can be difficult. And then to add to the complexity of sort of bringing genomics to the clinic in pediatric oncology are some aspects of the pediatric cancer genome. So I think all of you know and are familiar with this diagram from Bert Vogelstein's review and science and there are other pictures of this that show that the pediatric the cancers that tend to occur in the pediatric age range have a very low number of amino acid changing single nucleotide variants or indels. And so where adult cancers have sort of 1 to 10 mutations per megabase of DNA pediatric cancers even those with a high mutation rate have about 0.4 to 0.5 mutations per megabase of DNA. And some pediatric tumors have very few mutations and the best example of that is rabdoid tumors. However I would like to point out that it may be that other genomic mechanisms that are not detected by whole exome sequencing which is what most of these types of figures are based on may in fact be present and may in fact be targetable. So some have interpreted this lack of mutations to mean that there are more limited opportunities for practicing precision medicine in pediatric oncology. Others have said well if you find a mutation it's more likely to be a driver because there's less noise. So in fact pediatric oncology is the ideal circumstance in which to base therapy on the genomic mutations that you find. In fact we have taken a slightly different interpretation and that is really demonstrated by this figure which is from one of Mike Lawrence's recent pan cancer papers and that perspective is that we know very little about the genomes of most of the cancers that occur in the pediatric patient population. And basically what this figure shows is that the ability to detect recurrent mutations at the level of somewhere between 1 and 10 percent when you're doing a sort of discovery sequencing study depends on the number of tumor normal pairs sequence and the baseline somatic mutation frequency. And while pediatric oncology has a number of quiet tumors so the baseline or the background somatic mutation frequency is low our diseases are incredibly rare. And in fact the only disease in pediatric oncology that doesn't meet the NIH definition of a rare disease is leukemia and the only other disease that's been had a relatively large number of tumor normal pairs sequenced over a thousand is neuroblastoma. In fact, most of our cancers are like the one shown in this study and circled here in red where the number of tumor normal pairs sequenced to date is too few to say with any confidence that we've ruled out the possibility of a recurrent somatic mutation present at the frequency of 5 to 10 percent. In addition we have sequenced very few tumors from relapsed or refractory disease. And so really this is the simplistic way of saying that what we know about the pediatric cancer genome is really the tip of the iceberg and I promise that at the end of the talk I will come back to sort of how additional sequencing efforts discovery sequencing efforts like the target project and the pediatric cancer genome project led by St. Jude are hopefully going to help us move beyond this. But because we understand so little about the pediatric cancer genome the key question that we really wrestled with and continue to wrestle with is is it possible to extend the successes of precision medicine to pediatric patients with cancers where the key gene variants are not yet known. And at the time that we started wrestling with this question we were I was fortunate to be at the Dana-Farber and Boston Children's where an institute-wide project called the data file had been launched. This is an enterprise level research project between three hospitals Dana-Farber Cancer Institute Boston Children's and Brigham and Women's and it started in 2012 and essentially all children with cancer or suspected cancer seen at Boston Children's or Dana-Farber offered the opportunity to participate. And participation allows use of clinically acquired left over specimens for research sequencing for research which includes sequencing. We have pilot study in solid tumors where we have actually allocated some samples for creation of patient-drive models including cell lines and PDXs you can ask me questions about that at the end I won't touch on that more in this talk. It allows us to collect blood, cheek swabs and urine for research to use specimens and derivatives and place them in a tumor bank which is a virtual tumor bank so you can actually select the actual physical tumor bank and then the genomic data can be linked to clinical data for research purposes. Patients who have within this project or this study patients who have greater than 10 FFP unstained slides with 20% viable tumor have a sample submitted for tumor-only sequencing performed in a clinical lab and results with potential clinical impact with potential to impact clinical care or return to the primary oncologist and the patient. I will come back later to sort of where we are with this study but when this study started we sort of looked around and thought to ourselves if we do this at the Dana-Farber only we will never learn anything or at least it will be a long time until we learn anything from this project and I'm very impatient and so we actually launched a multi-institution study that's based loosely on the profile study. The sequencing test that is performed within profile and multi-institution study I'll tell you about is called Oncopanel. The sequencing success rate is very high it's 96% we use FFP as our sort of starting material the results are concordant with gold standard assays and it's both sensitive and precise. And it's performed in the center for advanced molecular diagnostics at the Brigham and Lemons hospital. There have been three versions of the Oncopanel test I'm showing you the 450 genes included in the most recent version these the top genes are sequenced for coding region changes and there are 60 genes where we capture across the introns so that we can identify translocation events. The report generated from the center for advanced molecular diagnostics includes a diagram of the copy number alterations the variants are tiered based on sort of loosely defined criteria of clinical action ability and because this is tumor only sequencing I just want to point out that if particularly if a patient is of a non-caucasian ethnic background there will be a large number of tier 4 variants which are primarily rare germline polymorphisms and then there can be interpretation provided for the variants and the extent of interpretation is variable depending on who's signing out the report. As I mentioned we decided that we would actually like in pediatric oncology to conduct this project as a multi-institution study and this is actually the first multi-institution precision cancer medicine study in pediatric oncology and we called it the individualized cancer therapy or ICAT-1 study and our goal was to determine whether it was feasible to identify key gene variants and make what we called individualized cancer therapy or ICAT recommendations using the clinical sequencing test that were available to us at the time. Patients were eligible if they had high risk extra cranial solid tumors and we allowed patients to enroll up to age 30. The tumor samples used for sequencing were as I mentioned sort of clinically acquired meaning they were sitting in the pathology department which most often means that they are paraffin embedded samples. We accepted samples from diagnosis relapse and preferred to have paired samples if possible. Tumor was subjected to mutation detection was performed using the oncopanel test and I forgot to mention it's a target capture followed by you know next generation sequencing using alumina. We did perform a race CGH when there was sufficient tumor material for additional assessment of copy number changes and then occasionally would do sort of follow up test if that was recommended by the expert panel. We had an expert panel or a molecular oncology tumor board that reviewed the results for clinical significance. We made an ICAT recommendation if there was a variant present that we thought was likely to alter protein function. Other people have called that pathogenic. We prefer to use the term altering protein function that we thought was that there was sufficient evidence to support that it might actually be involved with cancer related behavior. We needed to have we sort of tiered the ICAT recommendation depending on what evidence linked that variant to response to targeted therapy and we only made an ICAT recommendation if targeted therapy was actually available meaning that there was a dose in formulation that was appropriate for that patient. Our tiering system was we used tiers one and two. If there was clinical evidence supporting the link between the variant and response to targeted therapy, tiers three and four if it was preclinical evidence and tier five if we just thought it was a good idea because you always need that basket for consensus opinion. We issued what we called an ICAT report and sent that back to the patient via their oncologist. So in terms of results the first results that was really notable and was not known at the time was that there was a high degree of physician and patient engagement in a study like this. So that graph appears are the blue is our projected accrual and the red is our actual accrual. The second was that we actually could do this as a multi-institution effort in that 40% of the enrolled patients came from our three collaborating institutions and the first sort of take home knowledge was that 30% of the patients participating got what we called an ICAT recommendation and this has been published for some time now and you can look up what those recommendations were based on but loosely it was actual single nucleotide variants or indels, copy number alterations and then various rearrangement events. An additional 10% of patients had results that had implications for patient care and I'll go into that more in the next two slides and then we had someone from our population sciences department working with us who conducted surveys of the participants in this study and over 90% of the participants recognized that they were doing participating in this study to further our knowledge of cancer care which is how we explain the study to them and more than 90% said they would participate again. Okay so going into that additional 10% who had results with implications for patient care the first we found that 10% of the patients on tumor only sequencing had a somatic result that suggested the presence of a germline cancer predisposition syndrome but we did not do germline sequencing so it's actually better to turn to Will Parsons study which was one of the CSER funded projects called basic three where he actually enrolled a very similar patient population in that they had solid tumors so no hematologic malignancies but he enrolled patients at initial diagnosis and they did clinical sequencing using whole exome sequencing of tumor and germline and in his study he identified 10% of patients had true inherited cancer mutations those are the sort of known pathogenic mutations and this has been shown in subsequent studies as well and so an important lesson from these initial set of studies is that germline cancer predisposition is much more common in pediatric oncology than we previously appreciated so there were a few other patients who had interesting results that had implications for Karen and this slide really highlights that group of patients we did a pilot study where we selected nine cases for RNA sequencing and we based that selection on patients having undifferentiated sarcomas or translocation associated sarcomas where the translocation was not identified with standard testing in two of the patients the expected translocation was identified so that helped us to clarify diagnosis and in three patients translocations were identified that had either potential therapeutic or diagnostic implications for example a patient who was thought to have had Ewing sarcoma actually had a variant EWSR1 translocation that is associated with other diagnoses we had a patient who was diagnosed with melanoma who in fact had an EWSR1 ATF1 mutation which is much more consistent with the diagnosis of cutaneous clear cell sarcoma and then a patient who was called intermediate grade spindle cell sarcoma who because they had an ETV6 fish a break apart fish that was negative and this is the test that is typically done for infantile fibrosarcoma in fact they had an Ntrek3 fusion it was just a fusion with a different fusion partner EML4 and so in reality having an Ntrek3 fusion with an intermediate grade spindle cell sarcoma is infantile fibrosarcoma it's just that the standard test was unable to detect this novel fusion and so the lesson here is really two fold one is some of the genomic sequencing results you know we sort of got into this because we were interested in precision therapies but in fact we're starting to learn that genomics can help us with precision diagnosis but in addition we while some of these fusions could be detected by our current panel test some additional fusions that we found are unable to be detected that way and so the use of RNA sequencing in this way really highlighted the fact that it's really not yet known which tumor profiling assays optimally balance the competing factors of minimal tissue requirement comprehensive genomic assessment and rapid data analysis and results reporting so and then finally you know everyone's interested to know in our study how did the patients do right who got matched therapy right you have 30% of your patients who had an eye cat recommendation well it turns out that we had a relatively short follow up time of only six months and only three of the 31 patients who got eye cat recommendations actually got targeted therapy matched to the eye cat recommendation we did a little physician survey which of course had a low response rate asking why or why not did you give matched targeted therapy in some cases the expert panel had reviewed the case and you know theoretically a clinical trial was available but in reality for the patient that clinical trial wasn't available we didn't have access to sort of all the criteria for eligibility for that particular patient in other cases the clinical status was not what it needed to be in order for the patient to get therapy in some cases the patient was in second remission with standard relapse therapy and in some cases the disease was too advanced or the patient was deceased similar results were found in another clinical sequencing study that came out just before ours this was another NHGRI Caesar funded study conducted at the University of Michigan in which patients with all diagnoses were had clinical sequencing done using whole exome sequencing tumor normal pairs in this case as well a very small proportion of patients got matched targeted therapy and I think really we've learned in this field that this magic sweet spot that's required to receive matched targeted therapy where the gene variant is identified the patient has an appropriate clinical status and the drug is available is relatively rare and it's hard to make this happen and so we have we think that future studies in this space need to assess reasons for failure to receive matched targeted therapy and really importantly in pediatric oncology particularly we need more trials of targeted therapy where there are biomarker criteria for trial entry oh by the way none of the three patients responded we can talk about that more later if you want to ask me questions about that so I've told you about sort of the first three clinical sequencing studies that were published in pediatric oncology they all have very consistent results the sort of 30 percent rate of potentially targetable or potentially actionable variants germline cancer risk and this sense that there's probably more in the genome than what you can find by just looking for SMVs and indels there are additional clinical sequencing studies that have been published since then I just give you a list of those here but again the findings are very similar but we still have an answer question the first is our I should just emphasize here that the majority of these studies have primarily patients with either intracranial or extracranial solid tumors and the leukemias the feasibility of doing clinical sequencing returning results and looking for sort of match targeted therapy has been much not as well studied in the hematologic malignancies that's there I think there are a number of reasons for that some of it has to do with the pace of disease and relapse hematologic malignancies some of it has to do with the possibility of bone marrow transplant for those patients but I actually just want to point out that Minyon Lo and Kim Stegmeyer at UCSF and Dana Farber have collaborated with a number of other institutions to do sort of a very similar feasibility study that's focused on high risk leukemias and relapse leukemias and there's the clinical trial identifier for that the second unanswered question is what is the actual impact of receiving much targeted therapy on outcome so none of the sort of feasibility studies really answer that question they don't even we didn't even propose to try to look at that question in our ICAT-1 study the third I already mentioned what sequencing approach is really the optimal approach and finally I think we really don't yet understand the full spectrum of actionable variants that we see in our diseases so but what all of those clinical sequencing studies do is to provide us with enough evidence to say yeah we can do precision trials and pediatric oncology we think that's a good idea so I'll spend the next 20 minutes or so telling you about the precision trials that are ongoing in pediatric oncology and I'm gonna highlight three examples and I'll start with basket trials and recurrent disease and I'll start with the NCI Children's Oncology Group pediatric match and the first thing I like to say when I get to this slide it's very important to point out I am not one of the co-chairs of this study so if you don't like anything about it it's not my fault I am highly involved I'm the vice chair of the screening protocol and I am very involved in selecting study agents so basket studies are basically trial okay the term has been used loosely so I'll explain this basket trial but basically the idea is patients enter and in this case it's a relapse refractory trial and it's children are eligible if they have relapse refractory solid tumors or lymphomas and that includes histiocytosis which is important because those have activating mutations in the RAF MAP kinase pathway there needs to be some sort of tumor material that can be subjected to sequencing and in this case tumor material that's required needs to be from a relapse or refractory disease state it cannot be the newly diagnosed I'll talk a little bit more about that on the next slide the tumor is subjected to genetic sequencing and the way that this trial is designed is everything up until here where this little box drops down which is actually a targeted therapy trial arm is part of what's called the screening protocol so patient enrolls actually on the screening protocol we collect the tissue we do the sequencing we figure out whether or not there's an actual mutation detected and then we make a match to a clinical trial arm if one of the predetermined or pre-selected mutations is present so then there's a cassette of trial arms and for each trial arm of a targeted therapy the variance that would be allow a patient to get onto the trial are predetermined so this matching is done in a sort of automated fashion although there is human review before the patient is actually assigned to that arm of the trial then the patient is essentially on a phase 2 trial and the primary endpoint of each of those phase 2 trials is an objective response rate importantly some of the trial arms actually allow for enrollment of biomarker negative patients and that was done if there was a sense that there were biomarkers but that we didn't understand the full spectrum of the biomarkers that might predict response to a targeted therapy and I'll show you the example of that when we get to the slide that shows what drugs are actually part of this trial and then one final aspect of the design is that for each of these phase 2 trials of a study agent if there is some signal of activity in a particular histology so if I think it's 3 patients with a particular histology or on that arm and there is a signal of activity you can actually expand the trial for that particular histology so that you get a full phase 2 trial where the patients have a set of variance that predicts response plus a specific histology some more sort of detail about these specifics in terms of specimen requirements this trial actually allows for biopsy to be performed for study purposes if the biopsy is low risk and there's a whole description in the protocol of what that means and there's a good deal of evidence to suggest that complication rates from these low risk biopsies are very low in our patients and are not outside of the realm of the risks that our patients face in everyday care in terms of secondary objectives I just want to highlight a few one of them is that we do have a cell-free DNA objective and what we're really looking at here is the ability to use cell-free DNA to detect the genetic alterations that are present at the time a patient enrolls and then to use cell-free DNA to look for mechanisms of resistance there are three time points when cell-free DNA is collected we are interested in the patients will have germline mutation testing performed and it will be returned and that's because of the importance of germline mutation in these cancers and then finally I just want to mention that the match assay which many of you may be familiar with is using an ion torrent or hot technology and as I already mentioned the reporting which is sort of this automated reporting the results that get returned will either tell you that you match your patient matches to a trial arm and then here are some other variants that were found with minimal clinical interpretation of those additional variants so I am co-chair of what's called the target and agent prioritization committee with J. Cho our charge was to prioritize the most relevant molecular targets and corresponding agents to recommend to the trial leadership for consideration of inclusion as sort of trial arms the members of this committee represent this children's oncology group disease committees the NCI CTAP and NCI match which is we call adult match so match trial being done in adult patients and the FDA basically we went about this in a very sort of like you would a grant review we did a detailed assessment of all the target agent pairs that could be considered and we assigned those target agent pairs to reviewers who looked at all the literature we asked them to consider the following factors how common is the target in the eligible patient population what level of evidence is there linking that target to the agent activity or that mutation to the agent activity the minimum we required that there had to be at least some kind of clinical level of evidence so our minimum clinical level of evidence is a case report and a patient who has a variant in the target gene that's expected to alter protein function who has a response but with that because you can get that randomly right so occasionally patients respond to phase one trials you also have to have really good clinical data that demonstrates the mechanism or the sort of underlying biology that links the variant in that gene to response to that class of agents as I mentioned we sort of had this grand style review we discussed, we voted and then we had co-chair review we assessed the suitability of the match assay to identify the variants in the target gene or set of genes and also we looked at the impact of the trial design on priorities so importantly this mostly came up with the CDK46 inhibitors which were thought to be more cytostatic than cytotoxic and when you have an objective response rate as your primary endpoint you sort of have to think about whether or not you're going to expect objective response rate and then we made recommendations to the match leadership who then there's a sort of follow up process where you engage industry to actually select the agent from that class of agents so these are the initial arms of the NCI COG pediatric match protocol and we're showing you the agent class and the actual agent that's included in the trial and the example that I mentioned that allows biomarker negative patients are the PI3 kinase mTOR inhibitors because there are sort of signals that there might be predictive biomarkers but on the other hand there are certainly patients who respond to this class of agents without necessarily having a predictive biomarker so that basket trial like the NCI COG match trial is sort of what people typically think about when they are thinking about precision trial designs however in medical oncology there's been exploration of a lot of other precision medicine trial designs and I'm going to highlight two that are ongoing in pediatric oncology the first is sort of a real world evidence type of design and the second is a pragmatic trial and the two of them are sort of related to each other the first one is our follow up to the ICAT one study and the second is the American Society of Clinical Oncology what's called TAPER trial after finishing our feasibility study the ICAT one study really was meant to be a feasibility study as I mentioned we felt that there were a number of questions remaining to be answered and a number of investigators at other institutions fortunately agreed with us and we formed a 12 member consortium called the genomic assessment informs novel therapy or gain consortium and our first clinical sequencing cohort study is actively enrolling patients and it's called the gain consortium ICAT two study the title is too long to read so I'll just tell you about it patients are focused on the same patient population as we focused on in ICAT one patients have extra cranial solid tumors that are either high risk for relapse due to their initial presenting features recurrent or we've actually brought in our eligibility just a little bit for this trial to specifically target patients who have an unclear diagnosis after standard histopathology review and molecular testing and the reason we actually called out these patients in our eligibility criteria is we realize they actually fit into this high risk group we actually don't know their prognosis so it's hard for the investigator who's seeing that patient to say they're eligible based on we by the way we define this newly diagnosed high risk is expected overall survival of sorry two year event free survival of 50% or less so that we would include high risk neuroblastoma these patients who have an unclear diagnosis there's no way for the investigator or the clinician to actually decide what their prognosis is but we all have a sense that their prognosis probably isn't great right if we don't know what they have we don't know how to treat them and treatment is the number one prognostic factor so we specifically allow those patients to enroll in terms of tumor profiling we continue to perform primarily on co-panel as our targeted next generation sequence pan sequencing panel test it says here tumor plus normal we are still doing tumor only sequencing because the center for advanced molecular diagnostics timeline for initiating normal on co-panel testing has been longer than we anticipated but we actually have all of our normal samples banked and plan to send them for testing as soon as they're able to do that which they promise me will be quite soon and then we actually have a very complex diagram where we then select patients for either whole exome sequencing and or RNA sequencing depending on what the diagnosis is and just I would summarize that to say we tend to send rare tumor types for whole exome sequencing because we think there's still discovery potential to that approach and we tend to send patients where we think there might be fusions for RNA sequencing and certainly all of the patients with an unclear diagnosis have their tumor sent for RNA sequencing the targeted panel and the whole exome sequencing are done in a clinical lab the RNA sequencing right now is done as research both of the whole exome and RNA seek are done in collaboration with the Broad Institute and we're hoping RNA sequencing will transition to a clinical lab to the clinical side of the Broad Institute before too long one thing that's sort of changed over time in this protocol is that more and more of our consortium partners have their own clinical testing labs now that are doing targeted panel testing or they've developed a relationship with a commercial lab and so we have after much discussion decided that we never want to repeat a targeted next gen sequencing panel test with a similar with a test that covers essentially the same set of genes which many of these tests do and so we actually have modified the protocol so that rather than accepting a sample for testing we can accept a report and data so that we don't actually waste tissue on repeating something that's already been done a very important part of our project is the variant curation that I touched on in the beginning of the talk what is this variant what is it doing to the protein what do we think its role is in cancer and then what we call the clinical interpretation which is determining what sort of clinical significance that variant has for that patient and we make an ICAP recommendation very similar to what we did before continuing with our sort of tiering of the evidence and then what's different in this study is we do quite a bit of follow up data collection of the patient's vital status treatment and treatment responses we do have a self-re-DNA aspect of this project as well we're assessing barriers to receiving treated therapy we are planning to hopefully integrate this project into the collaboration we have with the Broad where we're trying to create patient derived cell lines I was mentioning to a couple of people that to be honest my least favorite part of this project is the piece the informatics infrastructure that's needed to support a project like this so when you're collecting samples from different institutions you're doing testing and you have to return test results you have to make sure you don't lose things and you have to make sure you know how long you've had something so that people are waiting so we learned with our first project that you really need good informatics support for that and that there aren't really products out there that you can just that are off the shelf products that can do this and so we've worked with the University of Chicago which is one of our collaborating centers somewhere here it is and they have a pediatric oncologist who runs an IT group to develop what we call our gain cast and sample tracking or gain cast system which allows us to track everything and return results and even more importantly we've worked with them to develop a knowledge base which is also what we use to create our clinical interpretation reports which we call gain reports and the idea here is really to decrease the burden of that somatic variant interpretation by being sure that we're holding our knowledge and making it easily accessible so we can re-access knowledge about variants and about their clinical meeting in our diseases there are a number of knowledge bases out there my cancer genome is an example it's probably the first one that existed but we don't find them to be that helpful for our cancer types they tend to focus on phase 2 and 3 clinical trial evidence and sort of FDA indicated therapies and we tend to operate mostly in the pre-clinical realm so we tend to find that we need our own knowledge base so our study activated in November of 2015 we were active for about a year with only the Dana-Farber as a site then we had a second site that activated and finally once we got all the lawyers actually working on our project we were able to activate all but the final remaining site and we're now accruing at the rate that we expected to accrue patients we have 200 patients enrolled this gets me back to the point I made in the very beginning pediatric cancers are a collection of rare diseases so these are enrolled patients broken down by diagnosis and you will see that osteosarcoma is the most common diagnosis but our second most common diagnosis is rare this is a collection of single cases of various cancer types and you'll see that not far down is non-rabdo soft tissue sarcoma which isn't actually a single diagnosis it's a collection of rare diagnoses we've completed 137 oncopanels 24 and 23 and RNA sequencing and then we validated two of our RNA sequencing results and we have about 59 cell-free DNA samples collected this is the obligatory slide of the single case which I won't dwell on but just to say we've seen some dramatic responses this is a novel BRAF fusion patient who received a MEK inhibitor which was seen in a neuroendocrine carcinoma again a rare tumor so how are we going to analyze this study what is our primary objective well we are we have a descriptive primary objective we are going to describe progression free survival and overall survival is really where we're going to focus comparing patients who received not matched and matched therapy this is very similar to what was done by the MD Anderson phase 1 program it's not definitive proof that precision medicine works it's really a way to sort of start to explore whether or not this approach makes sense and we think that doing this practicing we also can learn we have a very long list of secondary objectives where we can learn derive key lessons that can really inform future either basket trials or types of precision trial designs so to answer this question some of our patients need to receive matched therapy and you'll remember that on the first study only 3 out of 30 patients received matched targeted therapy so we're really interested in approaches to sort of improving access and that's where this American society of clinical oncology taper or targeted agent and profiling utilization registry study comes in this trial was started to address the problems of patients with advanced cancer having no standard treatment options having had a genomic profiling test performed having a potentially actionable alteration detected and having an FDA approved drug that was available but finding it difficult to get that drug and also losing the opportunity to learn if they did actually get access to that drug so I'll go through these taper slides quite quickly basically this study starts with somebody already having had molecular profiling done deciding to participate in the study if there's a predetermined match to one of the taper arms the patient just goes on that arm if not there is a molecular tumor board mechanism where the profiling result can be reviewed and then the pharmaceutical company actually provides the study drug at no cost to the patient and then the patient is followed for toxicity there are two sets of eligibility criteria one for the general protocol and then drug specific eligibility criteria but important to note that patients who are 12 and older are eligible and that was a sort of policy decision on ASCO's part and patients are eligible if they had solid tumors or B-cell non-Hodgkin lymphoma and the primary objective here is to describe anti-tumor activity and toxicity of commercially available targeted anti-cancer drugs in patients whose tumors have a genomic variant known to be a drug target or thought to predict sensitivity to a drug and the primary endpoint is objective response rate or stable disease at 16 weeks and the sort of statistical analysis plan is to enroll 10 patients per group and the groups are determined by tumor type variant and drug interestingly they're experiencing a very similar very long tail of rare diagnoses that we are and so there's probably going to be some modifications to this plan to do type variant drug they're probably going to have to lump some of the tumor types together the drugs will look very familiar to you there are only so many truly targeted therapies in the world and so many of these precision trials are using overlapping drugs and then so that really concludes the precision trials that I was going to go into in detail but I wanted to point out that there are many ways to conduct research in precision oncology and so there are other basket trials in relapse disease there are disease specific basket trials there are precision trials in newly diagnosed therapy in newly diagnosed parent patients which are primarily in pediatric oncology being conducted in leukemia and there are early phase targeted therapy trials in biomarker positive patients like that Lerotrectinib trial I showed you the image from in the very beginning of the talk and then we spoke about these in detail so in the last two minutes I just want to put a plug in for data aggregation and big data this is not my field so please don't ask me hard questions about this I just want to point out that there are several efforts to aggregate pediatric sequencing data each of them has their own kind of their own approach and their own kind of sweet spot so there's the genomic data commons which is an NIH funded effort which is really focused on the target data set which is the sort of TCGA of pediatric cancers and they are accessing primarily data from our cooperative group trials and are really working on standardization of clinical data elements St. Jude pulls any pediatric data that gets put anywhere into something that they call PCAN which stands for pediatric cancer their ability to achieve clinical annotation is a little bit less because they have incredible computational biology sort of acumen in the pediatric oncology space Gini which is the one that I most involve with is a project that's funded by ACR that's really trying to get clinical sequencing labs that are in hospitals to deposit data and aggregate data it tends to be focused on targeted gene panels the sort of sweet spot here is that there's incredible potential for clinical annotation because the data lives in a hospital space where we have the entire medical record to pull from now how you exactly pull meaningful clinical data from the medical record and annotate it with the genomics is another topic and then finally foundation medicine has actually made its pediatric sequencing data publicly available and I'll just show you a case study using some aggregated data so in the NCI COG match trial I mentioned that we looked for frequent like we wanted variants or targets that were had frequent alterations well we've sort of mined all of those and we have arms for all of those so we're starting to look at what are some of the rare variants that we might see now that we have more sequencing data available to us in these aggregated spaces when you look at that you actually see that there are EGFR mutations in pediatric cancers and MDM 2 amplification by the way this is from 981 cases where the diagnosis would make them eligible for the NCI match trial and the MDM 2 amplification is present at about 2% I'm going back to our profile or institutional study we now have 1200 patients enrolled and about 650 patients have ankle panel results available and we're able to mine those as well and when you look there again you see about one and a half percent of patients with MDM 2 amplification and so you can begin to think about whether or not you want to create another arm of the pediatric match trial that has the ability for these rarer mutation patients to enroll so in conclusion I would say that with the current technologies and understanding of genomics and complement of targeted therapies available many children with cancer still do not benefit from a precision medicine approach but when all aspects of the precision oncology paradigm align clinical impact can be significant for individual patients we have now completed studies demonstrating feasibility and pediatric precision cancer the pediatric precision cancer medicine field is transitioning to evaluating clinical outcome through precision trials and I believe that the number of children with cancer benefitting from a precision oncology approach will increase we'll need continued sequencing particularly of our rarer diagnoses that means anything other than leukemia and neuroblastoma and we'll need to use multiple approaches and sequencing so we can be sure to detect the variants that are not SMVs and indels we will need multiple precision trials with various and innovative designs and we definitely need data sharing and most importantly analysis of that data and so we need more computational biologists who care about pediatric oncology okay because this is the Trent lecture I'm doing an advertisement for Dr. Trent if you enjoyed this talk which I hope you did you can learn more about pediatric precision cancer medicine at a conference Dr. Kahn and Dr. Trent are leading which is going to be in March in Scottsdale, Arizona and it takes a village to do this kind of work and this is my very long acknowledgement slide with my very nice research team pictured down here on the bottom thank you very much for your attention well thank you very much Catherine for that terrific talk that was really fantastic I have just a small token of our appreciation of your giving the Trent lecture this afternoon for you to take with you. Thank you very much and maybe I'll start out the questioning since you wanted to avoid questions on big data and aggregation I'll start there with the interrogation but maybe an easier question it would seem like you know using the analogy of the tip of the iceberg there probably are lots and lots of cases out there where sequencing has been done in the United States but where those data have not yet been incorporated into this analysis. What do you think the magnitude of that is and what are the opportunities even internationally to increase the number since it clearly is the case that this is a matter of numbers to help things along. Yeah that's a great question I it's interesting I mean there aren't that many places that are doing the kind of sequencing that is high enough quality that we really want to see it so or that we want to use it for analysis so I think that the majority of those places are depositing their data somewhere the question is can you get it all in one place and can you stimulate sort of computational biologists to really work with that data and discover now I think that's true in the U.S. internationally is another question Europe has a very cooperative approach to this and they will have plans for data aggregation where I think that we have not made great efforts or at least I am not aware of great efforts is sort of extending beyond Europe and there I think there's probably potential for aggregation. Of course there's huge capacity sequencing capacity in China. Yeah exactly. Where one would think that a lot of opportunities. I think one of the challenges or one of the hurdles has been concerns about consent and that I think is actually an informatics problem that can be solved at least that's what I'm told. Why don't we open things up for others who may have questions as well are there any other questions from the group. Well if not we have a reception outside and you can direct your questions to Dr. Jane Way at the reception. So anyway thank you again very much for being here.