>>First let me thank the brave souls who are here, who crossed the frozen tundra to come to this really amazing lecture that we're going to hear today. I also know that there are lots of people online and I apologize for the late start, but we didn't want people to rush here because it is slippery still on campus. Our speaker today is Sam Mbulaiteye, and Sam is a senior investigator in the Infections and Immunoepidemiology Branch in the Division of Cancer Epidemiology and Genetics at NCI. Sam got his medical degree from Makerere University in Kampala, Uganda, and got his Ph.D. in epidemiology and biostatistics at Cambridge University in England. He went back to Uganda, and did his further medical training at Makerere University, and then joined the Uganda Cancer Institute and began his studies of interactions of the environment with human disease studying human immunodeficiency virus -- HIV and the effect of that on cancer in both children and adults. He then moved to the Uganda Virus Research Institute and following that he came to the NIH to work as a research fellow in DCEG. In DCEG, his work has taken a few different turns, although he still maintained his interest in the epidemiology of HIV and its effect on human cancer. And he studied two major issues, I think his great love is to study Burkitt lymphoma and understand the interplay of factors that lead to the high incidence of this disease in children in Africa. And I think he's going to be focusing a little bit on that in his talk. He's also interested in Kaposi sarcoma, which again, is an interesting disease in terms of the fact that not everyone with HIV gets this disease, there's obviously a role for herpes virus 8 and other environmental factors. So I think without further adieu, Sam, let's begin your talk entitled "Burkitt lymphoma: A model of polymicrobial carcinogenesis and global oncology. [applause] >>Thank you, Dr. Gottesman. I would like to express my absolute pleasure for you inviting me here to share about my work. It's really a unique honor for me, and I appreciate the opportunity, not only to present my work, but also receive the feedback that will be coming from all the listeners here as well as those who are watching the videocast. As you'll see, some of my work will appear preliminary, but that's because the field of Burkitt lymphoma research stagnated after going through ten years of really vibrant, exciting research, there was stagnation until recently, when studies have been initiated toward outstanding questions. My talk will cover some historical perspectives, just to bring everybody to the same page, but I will mostly touch on malaria and EBV, as etiological factors, specifically reviewing the evidence for association and also describe some of the ideas that I have borrowed from malaria vaccine and diagnostic studies to expand the knowledge of what we have about associations with BL, discuss the potential for proteome and genomics technologies to share answers to the outstanding questions, and then I hope to just illustrate how BL starts as a model for international as well as multi-disciplinary collaboration, as well as some trans-NCI and trans-NIH collaborations. Throughout my talk, I hope you'll really appreciate that BL is a complex model of carcinogenesis but I would like to first say that I am very pleased that the NCI has placed priority on Burkitt lymphoma -- my invitation to speak here speaks to that. But this point was made amply by Dr. Varmus when he made comments to the participants of the Global Health Burkitt lymphoma Meeting that was organized by the NCI Center for Global Health in 2012. First, the story begins with Denis Burkitt. And I'm really very excited to talk about it because Burkitt lymphoma is probably the most amazing story that's come out of Africa. This is Denis Burkitt. He was an Irish surgeon with one eye but an extraordinary vision. And part of what I will describe will talk about this extraordinary vision. It was his work in Africa, particularly his ability to convince people to get interested in obscure African tumors, that really unleashed the tidal wave of discoveries that followed in tumor virology, tumor biology, cellular biology, immunology, as well as international collaboration. When you think about Burkitt lymphoma, most people think about a child with a large facial tumor that distorts his appearance, but we do know that Burkitt lymphoma is a B-cell lymphoma, that occurs as three subtypes - endemic, sporadic, and immunodeficiency-associated. We know it's the commonest childhood cancer in equatorial Africa, where malaria occurs. We also know it's a rapidly growing tumor, which suggested right from the beginning that it might have a short latency, as well as it could be curable. It is fatal without treatment. However, if you have treatment, it's probably one of those cancers that you'd wish a child gets because the outcomes usually are very good. Burkitt lymphoma -- Denis Burkitt's seminal contribution to Burkitt lymphoma really began with that map -- describing the epidemiology of Burkitt lymphoma in Africa. And what he found was the disease was common in areas where mosquito-borne diseases were common. And that led to the first hypothesis that there might be a viral factor that was being transmitted by mosquitos that was causing this disease. This attracted the interest of Epstein, and Epstein was able to convince the medical research council to give him a small grant for virus discovery in cancer tissues from Africa. That led to the discovery of Epstein-Barr virus, which we now know is the first human virus to be linked with cancer, and has been linked to multiple cancers since, as a carcinogen. However, studies quickly showed that EBV was ubiquitous and it was not vector borne, and that observation led Burkitt and his colleagues who were working in Africa to, in fact, reformulate their hypothesis, that the vector-borne disease was not EBV but malaria which was cooperating with EBV to cause Burkitt. Denis Burkitt was a surgeon, and I'm pretty sure he was frustrated that his surgical techniques could not work on Burkitt lymphoma kids. So he was really willing to try anything to help these kids. And he accepted to test chemotherapy, and was the first to demonstrate that, in fact, cases could respond dramatically and some of them were cured when he applied his single chemotherapy. There was a lot of interest in the findings that Denis Burkitt was describing along with other scientists who were working with this issue, and those findings attracted the interest of scientists here at the NCI in the intramural program, which culminated in establishing the Uganda Cancer Center, which was led initially by John Ziegler in 1967 in Kampala. John went to Uganda and shaped this center into a world-class center. He began training local doctors to become research scientists and in fact, the work was so robust that many, many years later when I joined the medical school, I was also taken up into that field and became a cancer epidemiologist. I had the pleasure of working with John in the 1990s when he came back to Uganda. But there were also interests in the epidemiology program and here I'm showing work that was done at Korle Bu University, led by investigators from DCEG -- Bob Biggar and Paul Levine. And they went to Ghana to establish a field epidemiology program which I have benefited from. I feel very fortunate because I had the chance to work in the Uganda program as well as the Ghana program. However, let me go back to why I've been very interested in this condition. When I was training as an internal medicine specialist in Uganda, I was very interested in initiating my research career in Burkitt lymphoma, so I went back and reviewed the evidence that we had about the link between Burkitt lymphoma and malaria. And I started with Harrison's "Principles of Internal Medicine", and I'm showing here an example of the 15th edition. What I found astonished me. It was just two sentences in a chapter of ten pages, in a book of more than 2000 pages. And basically all it said was "it's possible that malaria is related -- malaria-related immunosuppression provokes infection with lymphoma viruses. Burkitt lymphoma is strongly associated with Epstein-Barr virus. The prevalence of this childhood tumor is high in malaria areas in Africa." Just that. For this [team of which] was so amazing, that he [had to go to the NCI to go and establish two research centers in Africa. The main limitations of the data that I could find, either in Harrison's or in the literature that I reviewed was that there was no information about the direct causation of malaria and Burkitt lymphoma. In particular there was no information on the magnitude of risk, if one was experiencing bouts of malaria episodes, there was no information about the specific malaria proteins that were involved in Burkitt lymphoma, or EBV proteins, or the role of genetic diversity of the pathogens -- EBV or malaria -- in influencing the risk for Burkitt lymphoma. And there was no information about genetic risk factors for Burkitt lymphoma. Clearly, when I look at the evidence closely I realize that most of the statements being made about the link between malaria and Burkitt lymphoma was based on ecological studies. And I'm showing here one example from 1963, just five years after Burkitt defined -- described Burkitt lymphoma. And you can see that the distribution of Burkitt lymphoma which is shown as dots corresponded to the distribution of malaria in Africa. This evidence has been repeatedly shown in different studies looking at this relationship at different scales. And I'm showing one example from Kenya in 2007, and you can see very clearly that areas where Burkitt lymphoma incidence is high, are the same areas where malaria incidence is high. This type of evidence is limited. It is inadequate for causal inference and doesn't tell us any information about the proteins that might be relevant for this relationship. So that's where I -- about the time when I met John Ziegler in Uganda doing his study of HIV and cancer in adults, and we decided to add a pediatric component to that so we could study 1) the impact of HIV on Burkitt lymphoma and other pediatric cancers, but I was particularly interested in the relationship between malaria and Burkitt lymphoma. John was working with colleagues from Oxford and the CDC, and together we were able to test patients that we enrolled at Mulago hospital for this whole scizont extract as a proxy for malaria exposure. This slide shows our results from that study that looked at 126 cases and 70 controls. Basically on the y axis, I show the antibody titers to the whole schizont extract. And on the x axis, I show the odds ratios. The one line shows a point of no association. And if the [inaudible] are on the left side, that means decreased risk, and if it's on the right side, that is increased risk. So we clearly showed that in these children, having high antibody titers for the whole schizont extract was strongly associated with Burkitt lymphoma. My colleagues from Oxford and the CDC went to Malawi, and essentially replicated our findings with an observation that kids with Burkitt lymphoma were more likely to have whole schizont antibodies in their system. These results were quite insightful but in fact, that was the entire world literature on the association between malaria and Burkitt lymphoma, and it confirmed the link but it did not answer the questions. For example, one question that wasn't answered was, since we know that being exposed to malaria triggers immune response resulting in clinical immunity to malaria and in some cases parasite immunity, does immunity to malaria protect against BL? That question was not answered by the two studies I just reviewed. So when I came to the NIH, I started digging into this question. My limitations included the fact that I could not pick from the shelf antigens or candidates that were well accepted by everybody as contributing to malaria protection. If there was a vaccine I would have just picked the vaccine but I actually had to go through the vaccine candidates and I picked SE36. I was very interested in that one because there was some evidence from people who were evaluating it as a vaccine candidate, that it does block parasite growth in vivo and in vitro. There was evidence from Uganda and elsewhere that showed that people who had high antibodies to SE36 were protected from severe malaria. The studies from Uganda were impressive for me because they were done in an area where malaria prevalence is so high, and the mosquito bite rates are so high -- the highest that have been recorded in the world -- 1500 bites per year or five infectious bites per night. So, I thought this was quite impressive. And what really made me focus on this antigen was the idea that people who were studying it were interested in collaborating with me on my questions in BL. So we put together a panel using data from the NCI Ghana Burkitt tumor project that I spoke about. This biobank has samples from 1965 to 1994 -- old samples but quite useful for my case. And we had 657 cases that were mostly strictly confirmed, and 498 controls that we studied. And the samples were tested by Professor Horri who studies SE36 in Japan. This slide summarizes our main results. You can see that these are the schizont titers, these are the odds ratios and these are the 95% confidence intervals. And what we found was, compared to those who were high, kids with Burkitt lymphoma were more likely to have low antibody titers, showing an inverse association. I was very excited about these results but people pointed out that "you only had one test. You did not have an antibody that is compatible with the exposure." So that was a limitation. And therefore I went back to the drawing board and decided to now pick two vaccine antigens -- I added MSP1, which was being tested, and then also added two diagnostic antigens - HRP-II, which is used to measure the pattern of malaria or to do routine diagnosis of malaria, and 6NANP, which is used by people who are testing CSP-based vaccines to determine whether somebody has had [inaudible] infection or not. This slide summarizes our main results for SE36. I should add that this time I changed my test lab from Japan to George Washington University, because I wanted to replicate the SE36 findings in a different lab. So I worked with a colleague at George Washington, Jeffrey Bethony. And basically, we showed that kids who had antibodies to SE36 had lower risk for Burkitt lymphoma -- confirming our initial observation. However, what was even more interesting was, in this set, where we now had data on another antigen, HRP-II, we found that kids who had antibodies to this antigen had a higher risk for Burkitt lymphoma. And when we combined the two tests and categorized individuals as either "not exposed/protected" or "exposed/not protected", then that group had 18 times the risk of Burkitt lymphoma compared to those who are in this group. And I began realizing right there and then, that we need to use multiple antigens to evaluate malaria exposure. We agree on the association, but the stronger effects depend on which antigens you evaluated. We think there are two types of responses have been conveyed. One is a risk response which is by antigens that [enlist] antibodies that are irrelevant for protection. And probably protection that is generated by antigens that [enlist] protective immunity. My only limitation here now was I only had one antigen that I had demonstrated protection for. For HRP-II, at least I could count on the whole schizont assay as a second antigen, so I decided to select another antigen. And I went to VAR2CSA. This is an interesting choice, because VAR2CSA has been linked to placental malaria. Placental malaria occurs due to sequestration of malarial parasites in the placenta, and it causes intrauterine growth with [inaudible], abortion, as well as maternal anemia and sometimes death. However, studies have shown very clearly that antibodies to VAR2CSA block the sequestration process and prevent placental malaria. What was interesting for me, there was literature suggesting that you can find antibodies to VAR2CSA in men as well as in children, suggesting [inaudible] and I was very interested in finding out whether it might have a link with Burkitt lymphoma. And like [inaudible] or SE-36, I decided to collaborate with someone who studies VAR2CSA on a daily basis -- Professor Ali Salanti in Denmark. This slide summarizes our results from the study that we did with him. We looked at 121 cases from the Ghana Biobank, and 101 controls and we decided to test the samples using three technologies, an Elisa-based technology, a flow cytometry-based technology, and a Luminex-based technology. And what we found was, for all the tests we had exactly the same result, a protective effect, suggesting that these results are likely valid and probably do suggest a mechanism. We know how VAR2CSA works, but we don't know how it works in the case of BL. This was a new finding that was totally unexpected for us. We think that maybe BL tumor cells might express CSA. That's what my colleagues suggested to me. And if that was the case, then VAR2CSA positive parasites would adhere in the tumor leading to a weak antibody response. We thought that this one is an issue that we could test, and we did that by looking in our tumors. These are representative samples from 40 tumors that we had from our biobank, and we tested them by immunohistochemistry to check to see if VA2CSA can be detected in tumor cells. You can see an example of high expression, but it's not in the tumor cells. It is in the stroma. This is low expression, but again in the stroma. So this suggested to us that VAR2CSA might exist in these kids who have BL, but it's not being produced by the tumor. So the low antibody response is not tumor-induced but is probably inherent in the innate immune response by these kids. Our follow-up is to test the samples that we have for CSA and probably other glycans, [inaudible] sulphates, to see if they are freely circulating in the blood. And this is really making us think in a totally different way about how malaria might be interacting with these kids to induce Burkitt lymphoma. So I've showed you some results from studies where I kind of cherry-picked antigens. And as I was doing this, I realized that probably that's not the most efficient way to do this. The best way would be to use Pf arrays, scan thousands of peptides and see which ones are interesting in the positive direction and in the negative direction. And what I'm going to do is show you results that I have from what with Denise Doolan at Queensland Medical Research Institute in Australia, to use Pf arrays to test the feasibility of using them to discover interesting or relevant peptides. This slide summarizes results that we have generated by probing serum of 32 cases and 22 controls, and what you can see very, very clearly - here we have the columns representing subjects and the rows are representing reactivities for the different antigens. And what you can see very clearly is that our cases and controls are very similar in their reactivity. And this is reassuring, because we matched them on area. They also have similar magnitude of antibody responses, as well as rate of antibody responses. These results are important for two reasons. First of all, they show that you can use an array biomarker to confirm similarity of exposure. This is important in case control studies where you are never sure whether you have matched your cases and controls well on a key exposure. For malaria, you could have to use a biomarker like this to confirm that similarity. Secondly, we did an analysis in a restricted set of peptides that were immunodominant, and we found that a small number were significantly associated with Burkitt lymphoma, 15. They had odds ratios ranging from 2 to 7, so suggesting there might be a way of discovering peptides that are important and relevant to this process. This work is in very early stages, and we are planning to expand it to use the biomarker for matching cases and controls, but also to scan additional cases so that we can generate hypotheses as well as test the possibility of getting a biomarker developed, a risk biomarker for Burkitt lymphoma. So up to this point, my experiences studying Burkitt lymphoma have showed me what the real challenges are. I don't think the challenges are those of ideas - what is relevant and interesting to study. The challenges have been really about sparse data that's out there to generate new directions, because there have been few studies, or the studies that have been there have few subjects that have been included and in my case, most of the samples that I have been using are rapidly diminishing. The Ghana Biobank will soon have no samples left. So if I have an idea I wouldn't be able to test it. With risk factor data we had difficulty, imprecise exposure measurement -- which exposure should you use and also a lack of pathology (diagnostic misclassification), -- all those are inherent challenges that we've had to deal with. So as I was thinking about that I thought it was time to renew this biobank, so that we could answer the questions that we had in mind. And I teamed up with investigators at the NIH, one of them is Dr. Bhatia, who always helps me think through the EBV questions as well as malaria immunology questions, Dr. Goedert who is here, he's the gentleman who recruited me when I was still in Uganda, and invited me to work here at the NIH - I appreciate that, Ruth Pfeiffer who brings her expertise, and Dr. Chanock, who brings all the population genetics expertise that I need. I invited him to join my team before he became my Director, but I appreciate the fact that he continues to show an interest. So the study that we are doing is being conducted in east Africa in three countries, these are the areas where we are conducting the study. And basically our plan is to enroll 1500 BL cases, who are aged less than 16 years, from defined geographical regions, histologically confirmed, and we collect clinical and demographical data. And from the same areas, we are interested in getting 3000 representative matched controls of the same age, same area, but do not have BL and we collect extensive specimens as well as questionnaire data -- large saliva, stool, tumor tissue, and questionnaire data on a range of risk factors. Working with this study has given me very interesting insights on what I couldn't do before. And I've just -- I'm just going to show one example. The question that I had in mind was - if you look at a child who has Burkitt lymphoma, when they have just come to you, are they more or less likely to have current malaria, if they live in an area where malaria transmission is so high? In other words, do they have worse or better immunity for malaria than age-matched children? We are able to look at this question because we are measuring malaria positivity in the cases and controls, using the thin, the thick and the thin film. These are examples of positive slides. And we also use a rapid diagnostic test, the HRP based, which I mentioned. Here is an example of someone who has complex infection, one and two positive strips there, and this is just simple infection and this is a negative test. So using these two tests you can actually determine whether somebody has current infection or recent infection because if they have recent infection but not current, they will be negative on their microscopy result, but positive on this one, so you can ask the question that I posed for myself. My results have been very surprising. The first thing we found was kids with Burkitt lymphoma are less likely to have current or recent malaria, compared to controls who come from the same area. The risk is much less as you can see there. If you look at the amount of parasites in blood, it's the same thing. You find lower parisitemia rates compared to controls. The controls are healthy children who we enrolled from their homes. My first thought was that maybe these kids with Burkitt lymphoma have been treated for malaria, so we asked them questions about outpatient treatment for malaria. This is mild treatment. And what we found again was if you look in the past 12 months, kids with Burkitt lymphoma are less likely to report a history of treatment for malaria as an outpatient. And we also asked them about severe malaria, are they likely to be treated, and we found again that kids with Burkitt lymphoma are less likely to be treated for severe malaria. And if you put these variables together and put other variables that we think might contribute to this relationship into the models, and this relationship remains, it seems that the association between current and recent malaria are independent of a history of recent treatment for malaria, either as an inpatient or an outpatient. And therefore they are not confounded by that, as you can see there. So my results that clinical malaria is lower in BL cases suggests that cases of BL have clinical immunity, and probably effective clearance of asexual parisites in the year preceding the time when they get Burkitt lymphoma. This was unexpected because malaria dogma implies that malaria immunosuppression increases EBV load. What we are finding is that in fact, the kids with Burkitt lymphoma have lower malaria risk. It seems to me that this is probably part of a Darwinian process that we are observing, and I just wanted to lay out the facts. About 200 million cases of malaria occur globally, and about one million deaths occur. The numbers are slightly lower for death now. And about 6,000 BL kids occur, and they all die if you don't treat them. Previously, our thinking was that Burkitt lymphoma arises from this group which is vulnerable with malaria. My results are suggesting that, in fact, they probably arise from a different group that has resisted or has innate resistance to severe malaria mortality. The mechanisms might include, the way they handle glycans in their system, probably it's genetically controlled. But anyhow, I wanted to kind of link some of my findings to this publication. Investigators in DCEG in 2013 described a new gene called IFNL4. This gene has been selected in such a way that people who are in Africa do have the wild-type, but people who are in Europe and in Asia have lost the wild-type. They have a derived variant and having the derived variant results -- gives them an advantage in how they clear hepatitis C virus infection. What really interested me about this gene was the fact it has very different distribution in Africans and Caucasians, and it has some viral effects. And the people who discovered it, Dr. Prokunina-Olson here and Thomas O'Brien work in DCEG. So I wanted to see if there could be some BL effects with this gene. I'm going to summarize very briefly the results that we have on a very small panel of samples. What we found, this is the wild variant, this is the derived one. And when we looked at BL cases, the risk increases with each TT allele you have, in the recessive model you find that it increases risk, even in the dominant model. It does suggest to us that this could be a genetic factor that contributes probably through an innate mechanism to EBV to BL risk, and Dr. Olsson is working with others including Jeffrey Cohen to try and find out if there's an EBV angle to this gene that could explain some of these very interesting findings that we're having. I would now like to shift gears very briefly to talk about EBV variants. It's been speculated for a very long time that the distribution of Burkitt lymphoma could also be explained by EBV variants that are tumor specific. This idea was suggested by Kishor Bhatia here. And in 2009, we reviewed the evidence such as there is to support it, and we found there was no clear evidence to support it one way or the other. So we decided that we would look into this question because I had tumors from Ghana and I was about to get tumors that Kishor had worked on from Brazil, so look at EBV variants, in tumor EBV from two geographical areas. And basically the idea is you could actually find the sequences that are relevant if they are common in the two geographical areas, but you would be able to also separate them from sequences that are irrelevant but tracking the geography because they are in that geographical area but not relevant. This slide shows results from our four BL tumors from Brazil and five tumors from Ghana. And you can see that we were able to sequence full length virus from the tumors and you can see that we had about 27 to 90 EBV copies per cell, that we found. This work has been done by colleagues [inaudible], who have sequence capacity to do this and the bioinformatics to analyze the data. What was very interesting for us was when we did a full genome phylogenetic tree analysis, we found that all the BL cases clustered together, I should say 8 of 9, because there's one that clustered here. And they were distinct from the NPC that are published in the database, And also different from the health samples that are published in the database. This VGO is from Brazil and clusters with these individuals, so probably EBV from somebody from East Asia. We've looked at the clustering at different gene levels and here I'm showing results for LMP1. And what we found was interesting, again, they cluster together, the BL tumors from Brazil and Africa. This one remains an outlier, by that distinct from NPC and distinct from the health samples from individuals from U.S., Africa, and Brazil. When we look at where the variation occurs we found that in fact it is limited to just a few genes, about 21 genes. But in fact those variations are limited to probably two genes. These are variations in tumor genes. We are very excited about the results, they are very preliminary, but they do suggest two things. One is that the BL EBVs from two continents are distinct and similar in certain respects. They are distinct from NPC, they are distinct from healthy individuals -- EBV collected from healthy individuals. They have unique variations in EBNA and LMP. Unique or common to them. The LMP variations are new, they have not been described in the literature. The EBNA variations, some of them have been described in literature particularly in tumor viruses from Africa. So to start concluding, I hope that my talk has really shown you how the intramural program has offered me a wonderful opportunity to do my research on BL. The study that I'm doing is an intramural-led study, similar to the studies that were done 30-40 years ago from the intramural program in terms of starting from the NIH, but different in certain respects, because we focus on just malaria, EBV and genetic risk factors. We think we will be able to answer those questions. However, the study has also provided a platform for innovative collaborations, and I will just mention a couple of them. The first one is with Ludmila, which I mentioned before. But we also have a collaboration with Michael Lenardo to look at magnesium and EBV in kids who have BL. We are working on a BL molecular staging project with Dr. Warren from the Fred Hutch, and we have a project with Dr. Guro Lind Elizabeth from Oslo, to look at the epigenetics of Burkitt lymphoma. There have been other trans-NIH collaborations, and here I just want to acknowledge the contributions by Dr. Steve Reynolds, who works for Tom Quinn. He's been very helpful in my work, particularly with logistics, ethics, and shipment of samples as well as long-term storage in country. I'm hoping that this will expand to other NIAID investigators because they have more expertise in malaria. And then I want to highlight the collaboration we had with Julia Royall when she was at the NLM - that was with Health Communication. When we started doing our study, we did not have any information about how to talk about Burkitt lymphoma and NLM helped us with that, linking it to the malaria work as well. And then recently we've initiated collaboration with the Fogarty International Center for principal investigator training, because I have discovered that my colleagues from Africa require some support in terms of principal investigator responsibilities. My work with Fogarty started with Ellis McKenzie. I want to acknowledge him for telling me it was not right for me to do Burkitt lymphoma research in Africa without highlighting the terrible mortality for the kids who get Burkitt lymphoma in that region. Diagnosis is a problem; people who work on Burkitt lymphoma know that this is a problem. We're collaborating with CIT engineers, Tom Pohida and John Kareka to implement CRADA 02720, to develop a benchtop ultrasound-assisted formalin fixation and processing device that could be used in small centers. Particularly for [__pathological samples which don't fit well in the big tissue processors. We think that the work that we're doing particularly collecting very nice tumor tissues is going to be very useful for this project, which is led by the Center for Cancer Genomics here at NCI and might provide an initial entry for African cases in precision medicine. I would like to thank in a special way these individuals, Robert Biggar, Joseph Fraumeni and John Ziegler, because they pointed me in this direction and have been very supportive throughout my career. But I also would like to thank all those who helped me, especially my colleagues at the NIH. You can see my colleagues that I mentioned, Stephen Chanock is here, Louis Staudt - we've done some work with him, Kishor Bhatia, who is always advising me, and James Goedert, as well as these others. And with that, I would be pleased to accept questions. [applause] >> Questions? And I ask that the people asking questions go to the microphone because there's an audience listening in. >> So you showed very nicely some polymorphisms or changes in Epstein Barr virus nuclear antigen 1, EBNA1, in individuals with Burkitt lymphoma, and as you know, EBNA1 is the major protein that's expressed of all the EBV latency genes in Burkit lymphoma. And it's a target for cytotoxic T-cells. Do you know if the changes you saw in EBNA1 are changes in cytotoxic T-cells epitopes, and whether some of these changes you saw in the EBNA1 sequence might correlate with the immunoscape - so to speak - of Burkitt lymphoma, controlled by cytotoxic t-cells? >> No, we haven't done any functional analysis of this data. What I have now is just where the changes are occurring, and what was very exciting for us was to find these changes common in tumors from two different geographic areas. They're all tumor viruses, so they must have the signature that is relevant for carcinogenesis. We plan to do some additional analyses, myself, but mostly my colleagues at the F.D.A. and working with Kishor here who has a wealth of knowledge about EBV in BL tumors to try and understand what those changes mean. The way I arrived at this point was - I was trying to figure out how do you study the relevant EBV variants in an epidemiological setting, and there are two possibilities that we're thinking about. One is let's get EBV from healthy people, and find all the variants that are out there, find out how to measure them well, and then do a case-control sort of comparison. That approach has not worked very well. We've worked here and we didn't succeed easily, we were working with Paul [inaudible] on that, and then we said why don't we start with the tumors, because tumors actually are a sampling of EBV from those populations that have progressed to disease, so they should have the signatures that you want and if you can narrow down the regions that are important, then you can just focus on those regions by using PCR amplification and NGS sequencing to find the relevant variants and then those variants that are - come out in a case-control setting are the ones you would probably move to study for function. My studies have been limited when I think about function because I've been working - I say working alone, as an epidemiologist, as a clinician, but as I've moved out, I realize you cannot do good work that way. You end up with half the answer. So I'm very interested in expanding collaboration with people who do bench work, who can knock in, knock out genes and confirm which ones are relevant and which ones are irrelevant and how they are relevant, so maybe we will -- we should talk. We have samples that we can study to clarify those questions. >> Thanks. >> Steve? >> Thank you, Sam. I have a question about the age, what's really striking with the information that, you know, you provided about the recent malarial infections, what's really striking is that in this part of the world that it's a very restricted age in which you see the individuals who are hit with Burkitt lymphoma. You don't see it after adolescence per se, it's not very often even seen in people making it into adolescence. So now we're at -- what are your thoughts about the age as a reflection of developmental opportunities or developmental, you know, periods of time in which someone would be at risk for developing this. Have you started to look at, you know, or think about the question of the burden of malaria over time, that that may over time just help to get someone to an immunologic state where they aren't necessarily at risk for this? I mean, it's such a curious thing that it occurs in such a small age range, you know, when in fact malaria, you know, affects 400-500 million people across that belt. What are your thoughts? >>Right. So we are thinking about that. Our thoughts began with collecting good data and samples, so we have -- we asked many questions about fever, so I just presented the fevers that are malaria, but we also ask questions about fevers are not malaria, based on the report from the patient's caregiver. So we're able to really say that these findings are very specific to malaria, they are not, therefore, non-malaria fevers. We also have samples, we have blood clots, so we'll be able to look in that, quantify the malaria experience. But you're correct that it's striking. If malaria contributes -- and we know malaria doesn't go away in that region, people get bitten by mosquitoes every day -- why is it that the risk for Burkitt lymphoma peaks? It does suggest that there is either the developmental process that cuts down the risk process, and it might be driven by malaria immunity or might be driven by B-cell physiology, and I'm not sure which is which. Our kids are matched nicely for age, so we know that the results that we're finding in the cases should be similar to those in the controls, because they are all about the same age, they are all 7.4, 7.5 years, so they are old kids. These old kids - the cases have better immunity to malaria than the controls. So previously when I was thinking about how to design this study, I would have focused on the kids who are vulnerable to malaria, as the kids where you're likely to find the translocations, you know, if you could do that. Right now I think, no, you must study the kids who are hardy - those who are able to control the malaria infection. From the epidemiological perspective, if we can have a biomarker that allows us to separate these kids into phenotypes, the phenotypes that we've been thinking of are -- there's a phenotype that leads you to severe malaria, and your risk for death is very high if you're in that phenotype. There's a phenotype that leads you to asymptomatic malaria and your risk for death is low in that phenotype, but your risk for Burkitt lymphoma could be higher because you don't have overt clinical malaria but you do have submicroscopic malaria, that may still be driving the process. I hope that with our willingness to collaborate, you know, with different investigators who bring different strengths to the team, we should be able to generate the next steps, to answer the question that you asked. >> Thanks for an interesting talk. You mentioned you studied other fevers. Did you or anybody else find any association between the incidence or severity of malaria and the incidence or severity of other viral diseases such as chicken pox? >> No. Our samples are cross-sectional, and the questionnaire allows us to look back in the history of the kids' experiences. We did not ask specifically about chicken pox. Sorry. But we can't answer that question because we just used cross-sectional data. In the samples that we have, you could do a study by measuring for specific infections, and show if there's a correlation between malaria, for example, and those infections. You could also use data that we have on the history of malaria experience, and see if there's a correlation. What I have done, this was just to ask if you've had an experience of severe malaria in the past, let's say one year, are you different in terms of your current prevalence for malaria now than somebody who reports they've never had. And we found differences. We do find that kids who report they have had a previous episode of severe malaria are more likely to have current parasitaemia than those who don't and that's striking because we're talking about healthy children. I mean, not healthy -- but in the African context, they are in their homes playing and doing their stuff, so they are not sick. And they have parasitaemia. So it does seems like having severe malaria is a different phenotype from never having an episode of severe malaria. It has to be epidemiologically driven, I'm not sure exactly what the immunological processes are, but they are there. I've had some discussions with Sue Pierce and Louis Miller and Patrick Duffy. And their focus has always been on severe malaria. But it seems like We must also focus on kids who reside in areas where malaria is so prevalent, but never seem to suffer from it. >>Thank you very much for a very insightful talk. So the prevailing dogma - the work may well have been inspired by that - and that is geographic clustering of EBV and malaria parasite incidence, but your surprising findings of a relationship almost also underscores complex pathophysiology, and you hinted at the glycans, you know, parasite sequestration in placenta and of course, the correlates of that may well be at the molecular level. When you look at these kids, do you also find that there might be clustering at the level of other co-endemic parasites that could modify disease incidence - vis-a-vis the immunological paradigms that you are beginning to hint at in answer to one of the questions? So I'm quite interested in what do you think may well be going on, so if you subtract malaria only for a moment, and consider the other parasites in those geographical areas, whether you think you might arrive at correlates that might be even more informative in terms of what's really going on, at a pathophysiological level? >> Right. So we take stool samples and measure for stool parasites. We've not found appreciable levels of infestation for stool parasites to make any meaningful contribution to what we're seeing. We were very surprised about that. But then I realized that the immunization program - the expanded program of immunization, they give out Albendazole, or Albendazole as part of the treatment that kids get when they visit, so that has reduced parasite load. Maybe they do have lower levels, so we're not able to look at that, using the data that we already have. If we can find a good serological test we might look at some of the parasites to see if they do contribute at that level. The ones that we've looked at are EBV -- we've measured EBV in some of the cases, I didn't present those data. We did not find asymptomatic malaria to influence EBV levels. So the kids who are EBV - who are malaria positive and malaria negative don't have appreciably different EBV levels. But this is because we're looking at asymptomatic kids, and not kids who have severe or acute clinical malaria. Those tend to have different EBV levels. We're very interested in interaction with other pathogens, KSHV is one that I'm very interested in, because it occurs there. It influences the risk for Kaposi sarcoma. I'm interested in seeing how it interacts with malaria. There are some studies that are showing that there's an interaction between KSHV and malaria. So we'll need to see how KSHV, EBV, parasites and maybe other pathogens, you know, could contribute to this interaction, to either increase or decrease risk of a disease. >>Great talk, Sam. It just came to mind, do you know the BL tumors that you sequenced the EBV genome, did you look at the other genetic lesions that are also common in Burkitt lymphoma like the c-myc and were they all the same or were there any differences there? >> We haven't done those analyses yet. But we have the data. This is a project that is likely to grow. Ultimately, I would like to include the EMBLEM data in this project because we have much more granular data on EMBLEM than we have on the Ghana cases, for example. We don't have detailed questionnaire data, we don't have current malaria data on the Ghana cases so EMBLEM is going to allow us to ask some of those questions that you're alluding to. >> Sam , I have one question. You gave us the somewhat unexpected result that in children who get Burkitt lymphoma, there seems to be a decreased evidence of recent or current infection with malaria. You didn't expect that result. >> No. >> And you're pursuing what seems like a very reasonable approach, which is to look for genetic factors in the children that may account for their decreased incidence of malaria and so on. I guess another formal possibility is that acute malarial infection protects against Burkitt lymphoma, and one way you could check that as a control would be to look at other genetic factors that are known to protect against malaria, like sickle cell trait. So is there any association between Burkitt lymphoma and sickle cell trait? >> Yes, so we've looked in our samples, Miller here was able to test our samples for the sickle cell gene, and we find a non-significant protective effect. In fact, in the model for IFNL4, sickle cell trait does not confound that association and itself is not significant. So that's surprising to me. We're still exploring how these two fall together, sickle cell, IFNL4, and probably other genes that will be looked at. But what it really told me was I won't be able to get to the bottom of this question if I work alone. So I need to work with other people who bring different... >> There are beginning to be other genetic markers that people look at, that are protective. Malaria obviously is an enormously selective circumstance in African individuals, so genetics probably could give you some hints as to what's actually going on. >> Yeah. >> Any other questions? If not, let's thank Sam for a terrific talk. [applause] And I wish you a very careful walk back across the campus.