Colon Cancer Awareness Pins

today we have neil caporaso from the division of cancer genetics, way back in 1983 he came to nci as oncology fellow in the medicine branch. he stuck around all this time. he's chief of the genetic (inaudible) immunology branch atty grove facility where my

office is, the title of the talk is "epidemiology." >> today we have technical challenges. can you hear me? great. i have to simultaneously click both of these slide advancers, so this will strain the limits

of my coordination. let's hope it works out today. epidemiology is the most important lecture you'll get in this series because it enables you to prevent all the other cancers. i know we'll have fun. i'll tell you about basic

concepts, challenges and some ways the future of epidemiology will go. so we are part of nih, and under nih is the national cancer institute. my particular division which i'll tell you about in a second is one of the two intramural

divisions of nci. intramural is special, because we don't compete for grants exactly like the extramural world. most of you, most of the world, is out there writing grants and that's about 85% of the money, but in the intramural we get

money allocated, a certain amount of money, and we compete internally for that money. you say how come you're not competing too, we do compete internally and we're site visited every four years, it's a different competition. we're site visited and they

don't like us, our lab is shut down. the idea of the intramural division is you get to sustain projects that are not necessarily popular. we began doing family studies 30 years ago. at the time everyone thought

human genetics was a waste of time. really if you want to be genetics, you should be study, drosophila and why waste your time on humans, there have been a number of unpopular ideas that have been sustained over time and in fact they have paid off

somewhat. okay, this is dceg, and there are a number of epidemiology and genetics side, i'm on the genetics side, genetics of family. other groups look at hormones, disease, and we have multi-disciplinary teams,

physicians, oncologists, et cetera. our goal is to identify environmental and genetic causes of cancer in the population. a lot of public health advances have emerged from the work of our group and others, but specifically regulatory changes

in water, less benzene in gasoline, workplace safety, farming, it's kind of you throw up a line like this, workplace safety, diesel, the study of diesel exhaust was fought viciously by advocacy groups and it was an extremely difficult study to full off.

the occupational group was called before congress a few times. it was a difficult study in the public health so it was clear diesel exhaust is related to lung cancer. so there have been clinical studies of cancer susceptibility

syndromes and a lot of studies of second cancers. and different preventive interventions like how to do safer cat scans, risk-reducing surgeries, benefits of healthy weight, et cetera. and there are collaborations around the world.

i have a principal investigator that just came back from beijing, and saturday she left for kenya. so she has breast cancer studies in both places. and if you go on the web, you can find us and learn all about the kinds of studies that we do.

and if you go on the web you can tools. okay. so enough about our group. some basics of epidemiology, the term, epidemiology, comes from a study of people. in other words, populations. we don't study individuals,

clinical studies. we don't study cells, we don't study molecules, except as secondary goals. we study groups of people. and these are observational studies. so they are not experimental you can't assign people to smoke

cigarettes. that's not ethical. and that's a weakness of the epidemiology design, it's one that tobacco companies and other groups exploited to question the validity of epidemiologic findings. as i said, epidemiologists are

ethically prohibited from experimenting upon people. so we observe large populations, and essentially do studies that observe the outcomes in people and the exposures that people take on, and then do statistical analyses to try to determine the differences.

generally, the goals of epidemiology, particularly the goals, the central goal of our group, is to understand the causes of cancer. we also want to quantify risks, understand mechanisms, have implications for public health, and identify syndromes.

one of the features of epidemiology, we emphasize prevention. unlike clinical medicine, which looks at curing disease, which is a wonderful thing, i'm a physician, i'm totally in favor of it, but the idea of prevention is that it's very

effective. you think of vaccines. it's much less expensive. it's oriented towards public health and it eliminates disease early, before you have health consequences. but there are some real big problems to prevention.

the biggest problem is that it takes a lot of time to demonstrate that somethingworks. it's much less dramatic than treatment. you don't have grateful patients coming to you and saying you cured me. hey, no one's come to anybody

and said thank you for that vaccination. i'm glad i didn't get polio. but you should be glad. the other thing is that politically, when it comes time to allocate money, patients are much more effective in our political system than thanhealthy

populations. that's the way it goes. politicians tend to respond to dramatic cures, less so to just preventing disease. maybe that's not the reason exactly but you won't find many epidemiologists among those getting nobel prizes, so i think

personally that's a disgrace, but that's the way it works. epidemiologists worry about bias, which is a simple definition, systematic deviation from the truth. a lot of facts can make the facts skewed, a lot is participation.

who participates in a study? we want study subjects to be representative, the sample to be representative of the population from which it comes. and that's important if we want to generalize to the larger population. so, i did a lung cancer study in

italy, and the site visittist came. remember the site visitors i talked about? the study was going to cost about 12 million dollars. and so they reviewed the study and said, um, nice idea, this is the first -- the largest lung

cancer study in the world, one of the first studies to incorporate blood collection, also tissue collection. so by the way that tissue now is absolutely invaluable because fresh frozen tissue from the operating theater. anyway, they said to us, you

know, if you can't get your response rate over 50% so your lung cancer cases are representative of the general population, you can't do the study. we don't want that. so we did a phone survey where we called people up and said,

hey, how would you like to be part of a study in which we'll give you a questionnaire, it going to take 72 ccs blood, oh, yeah, you have to sign and informed consent the size of a telephone book. amazingly, only 30% agreed to do that.

we added invitation letters, phone follow-up, offered to study them in the hospital, put advertisements in the local newspapers, we gave them a cash award. we got a letter from their physician. and we still only got to 49%.

finally we got extremely charming interviewers of the opposite gender. we used a physician call. we gave them gas coupons which in italy was a big wonderful thing, 20 euro gas coupon, that was big. we had the local night show, the

11:30 night show, the equivalent of johnny carson at the time, have tv ads. we had a better invitation letter which was oriented towards a sixth grade education. we got a letter from the mayor, they liked the mayor in milan. and a toll-free telephone line.

and that got us to the participation rate. so it was not easy. when you read about participation rate, realize it takes a lot to do that, factually the series cost the first million dollars of the okay.

epidemiologists are obsessed with controls. population controls are thebest. in the olden days, we used to say that quality standard was random digit dialing. oh, did you do random digit trials to do your controls. today you get a 2% response

rate. but still, it's worth a struggle to get controls, and i don't want to go into all the details about controls but you really want your controls to be comparable to your cases or your study is really invalid and you don't want terrible controls,

convenience controls that are biased, by obvious differences in age, risk factors, ethnicity, education. so, you know, you don't want to go into your laboratory freezer and take out anything determined human. if you call an epidemiologist as

a consultant, and we love to consult and be party of studies run by other folks, they will ask what was your study design, where did the controls come from, did you collect key co-variant data, meaning factors we take into account in the analysis, how much did your

people smoke, how old were they, did they drink alcohol, did you consider bias as confounding, what was your original hypothesis, what you don't like to see is you do your study and then analyze every variable that you could possibly measure in the study for an association.

that's called data dredging. no, no. did you do power calculations to figure out how big your study has to be and validate your marker? and they be the most common question an epidemiologist gets

is, oh, my grandmother is 100, and she smokes, she drinks, all of her doctors have died. [ laughter ] she lives on bacon. you know, how do you explain that? explain that, explain my grandmother!

it's a probabilistic science. anything can happen to one people. in groups, the smokers are all going to die before the nonsmokers but there can alays be a smoker at the end of the gaussian distribution that lives a long time.

we want to study their genes. a study looked at 90-year-old smokers, very hard to find a lot of them but they did find a few. they claimed to find some genes that were associated with long life, i don't actually believe that study but okay. what kind of tools do

epidemiologists use? and one of the tools that we use are these cancer maps, these were made by the descriptive epidemiology folks in our division, a map of medical no, ma'am a melanoma. you take one look at the snap, look, they are getting more

melanoma in the south, okay? it's immediately suggesting sun has something to do with this. so these maps are very, very helpful. and you might say, well, we knew that already. what else can you do? and i'll show you in a second.

there are more advanced analyses of geographic data using gis systems. geographic information systems. and a lot of analyses are done with seer, the registry covers a quarter of the u.s. population, surveillance, epidemiology and end result, and this is an

organized registry of cancer data, and increasingly this data is getting better and better and it's being linked to other kinds of data. so seer-based studies are really important parts of cancer epidemiology and we're glad that reagan didn't stop this.

it's one of the areas when they were looking for areas to cut, he said what is this? why do we care? why is the government involved in this? well, seer, if you look at the studies that have come from seer, it's extraordinary how

much wonderful information we learned from seer. and it's a website that you can visit, and you can download data and make your own graphs and your own tables, if you're interested. and the kinds of things we can learn from seer, for instance

you look at this incidence rate in men, and the incidence rate in women, there's a bump around 1992 in men. you might say wow, what happened to the men there? what changed? you know, was there an epidemic of cancer?

does anyone know what caused well, the answer is it was due to psa screening for prostate. this was a bias in the data. it was a little blip that they introduced psa screening and a lot more people were identified with prostate cancer. it didn't necessarily affect

prostate mortality. it's not certain all those cancers were destined to progress, but it's one of the little features you have to take into account. so seer data gives you very good information on cancer rates by ethnicity and gender.

you can see in males, particularly black males, the bump in cancer, you don't see it in the women. difference between cancer incidence and mortality can inform lot of other things like the success of treatment. so the reason for this

difference in children is that we're very good at treating childhood cancers. if you do that for pancreas cancer, you don't see a big difference because we're not so good at treating pancreas cancer. well, one of the challenges to

epidemiologists is causation. how do you prove causation? because if we're working with statistics, what we get are statisticale associations, correlation doesn't prove causation, so epidemiologists have classic criteria to try to infer causality.

you don't prove causality from this but you can gather evidence in favor of it. the criteria are it's more likely to be causal if it's associated with a high risk, so a relative risk of 1.02, a 20% increased risk, 1.2, i'm sorry, not so impressive.

a relative risk of 10, 10-fold greater risk, a thousand percent, okay, that's much more impressive and much more likely to be causal. it should be consistent. we like to see it in different groups, in different populations, different

countries. we should see a dose response, if you increase the amount of exposure you want to see an increase generally in the amount of the cancer. you want to have something that makes sense temporally, the factor that you think causing

the cancer should come before the cancer. if it doesn't, that interferes with causality. finally, you'd like to have biological plausibility. i just want to mention that there are additional approaches that have been used more

recently in epidemiologic studies to infer causality. one is mendelian randomization, that's the idea that if you find that a series of genes are associated with an exposure, and those genes are also associated with cancer, it can weigh in, in the argument that that

particular factor is not just an association but a causal factor. molecular epidemiology refers to using biomarkers, allowing you to look at mechanism. mediation analysis is similar to mendelian randomization. this is the kind of data that was very compelling and proving

smoking was a true cause of lung cancer, and not as some famous scientists and a lot of industry shils argue, no, no, cigarettes have nothing to do with lung cancer, it's more risk-taking people smoke, they tend to get more cancer. so here you have cigarettes

smoked per day in three cohort studies, rates of lung cancer. there's a strong and consistent association. and here you have the temporal factor that the average numbers of cigarettes smoked followed, you had males smoking and females smoking and female lung

cancer rates going up, following the female smoking and male lung cancer rates following the male smoking. that's temporal. and then you have biological plausibility and this was the study of oscar auerbach who showed histologic changes were

similar to humans, a very compelling mechanistic and biological argument. finally another study from three large cohorts showed that when people stopped smoking, the rates of lung cancer over a 20-year period slowly returned almost to baseline, never quite,

but almost. the rates of myocardial infarction take six months to come down but prolonged cancer takes longer. so what are some aomplishments of epidemiology? and it looks like there were none.

oh. okay, so i didn't need to stuff this in but over last year there were 500 publications and a lot of press calls, and here is one on coffee drinking and total cancer mortality, i'm not going to talk about it. it got a lot of media calls.

actually i did a study with a large team a few years ago and we got hundreds of media calls identifying some of the genes associated with coffee drinking. here is another one on smoking and risk of bladder cancer among men and women. i don't want to go into these.

over the decades, there have been studies and a lot of controversy about silicone breast implants, chernobyl accident, oral cancer and mouthwash, abortion and breast cancer, cell phones and brain tumors, a lot of these were political.

in congress, someone would have a constituency like long island women that said we have to find out if all these chemicals in long island with causing us to get breast cancer and would mandate a certain study. other groups of republican politicians wanted to show

abortion was related to breast cancer which it wasn't. some of these studies were politically motivated but we're a government group and we have to respond if they tell to us do a study. the fukushima disaster groups from our radiation epidemiology

branch traveled there and helped a lot with both symmetry and following up on public health events. but i think that generally the most important accomplishments in epidemiology are that we've contributed to general an understanding of the general and

specific causes of cancer. we're advocates of public health. epidemiologists established through many, many studies and battles with industry and tobacco industry that tobacco is a causal factor for lung cancer, and that secondary tobacco smoke

was also a risk factor, which is what resulted in the clean air legislation. so every time you go in a restaurant or get in an airplane, there aren't people smoking next to you, you can thank epidemiologists. molecular epidemiology is the

idea that we can learn a lot more by including biospecimen collection and analyzing biospecimens and analyzing the chemicals, the patterns, the genes, to understand more about what causes disease. sy briefly, some highlights about general risk factors for

cancer, age is an important risk factor, generally environmental factors and genes and combinations of those are the general causes of cancer. and it's often divided into three groups, tobacco a third, diet a third, everything else a third.

we like to say that basically almost all cancer is due to the environment. and one of the lines of evidence in favor of this is that if you look at cancer incidence, according to the highest and lowest, that's the h and the l, the ratios are dramatically

different. so the rates of melanoma, much, much, much higher in australia than japan. now, those of you that are astute would say, wait a minute, the australians are anglo-saxon red-haired northern european extraction, much more sensitive

skin and the japanese are less so. and that's true. so genetic factors do contribute something. but most of those differences are environmental. here is an example of an environmental factor.

a province in china has the highest rates of lung cancer in the world. and if you go there this is what you find. tremendous amounts of indoor air pollution. a lot of air pollution in general in china.

when you take away that indoor pollution, part of what's happened there in the last ten to fifteen years is they have installed measures to control that pollution indoors, the rates of lung cancer and rates of copd have gone down. here is an example from one of

the cancer maps, and i don't know where the cancer map is. i thought i had it on the slide. there's an area in montana that has a little red dot, it turned out that that dot was from a copper smelter that had arsenic contamination. and so that was accelerating the

rates of lung cancer in that area. when you got rid of it, you got rid of the rates of lung cancer. tobacco is still a gigantic public health problem. you can read the statistics but the one that impressed me the most is that in the last

century, we had 100 million deaths from tobacco, that's a big number. but it's projected then in the next century we'll have a billion. that's ten-fold more. you might say how is that possible?

because adult smokers in the united states most recent data approaching 15%, and we used to be at 50%. the answer is it's not in the united states. it's in the developing world, it's in china and india. and the middle east and africa.

africa's, you know, a new and emerging market for cigarette companies. so we really have to do a lot more. it's mega death due to tobacco. and here is how the rates of smoking dropped in the united states.

since the surgeon general report. it takes decades for epidemiologic findings to reverberate and it takes a long time to correct incorrect findings. so, for example, you still see -- walk into the grocery

store and you still see low fat food as if low fat food was something that might be good for you. in fact, low fat food is a disaster. as you can see from the rising obesity rates in the united but it takes a long, long time

for these results to change. it will take a long, long time to change the dietary habits in the united states to affect healthy change. i need to give a nod to the folks who worked on environmental tobacco smoke because as i said earlier, it

was the discovery, the environmental tobacco smoke was definitely related to lung cancer and other smoking-related conditions, particularly in children, that allowed it to be classified as a class a carcinogen, and allowed clean air legislation that cleaned up

our buildings and our workplaces and our airplanes and our movie theaters and our restaurants. you notice this when you go to europe, or evain parts of the united states, where they are kind of lagging behind, that it's really a pleasure not to htte smoke around.

so i should at least mention that the number two carcinogen after alcohol, after tobacco, is alcohol. and associated with a number of major cancers. ionizing radiation, associated with leukemia, breast, lung, thyroid, head and neck cancer.

a lot of studies of ionizing radiation and cancer and our want to just touch on these. i'm not going to talk about them. chernobyl. non-ionizing radiation associated with all sorts of skin cancer.

tanning beds have managed to -- still exist, i don't know why, clearly associated with melanoma. but the effect of the sun that's more effective in getting people to protect is the aging effect. so infections in cancer are very, very important, and a

major victory for epidemiology was understanding that the association of cervical cancer with increased sexuality activity was actually due to the human papillomavirus and the fact now that we have a vaccine that you can use to prevent that virus is a gigantic advance.

and i'm not going to talk a lot more about these. except to say that study of the human microbiome allows us to detect and dissect the effects of infection not clinically apparent, an emerging important area that we'll hear more of over the next few years.

we don't think about occupational exposures a lot, but they are critical and a number of cohort studies in our group including studies of farmers have elucidated the effect of a number of herbicides and insecticides on a variety of human cancers.

in fact a study came out this week of the ranch hands, which are the vietnam veterans that handled the herbicides in vietnam that were contaminated with dioxin, and they showed that the rate of a precursor of multi myeloma was two to three times higher in folks exposed to

dioxin, m-gus is a universal precursor of myeloma, showing a particular risk factor for myeloma. here is the diesel exhaust study i mentioned earlier. and that after taking into account smoking and other risk factors they showed diesel

exhaust was a major cause of what are the challenges in epidemiology? well, there are some big gaps. among them are the fact that there's some cancers for which we don't know that much about the cause. understand what are the

extrinsic environmental risk factors. we also don't know much about gene environment. so there's a lot of publicity about gene environment, and gene environment is theoretically important, since we know that broken genes are part of the

origin of cancers, and so the environment must act in some way on those genes and we understand for a few cancers, but gene environment interaction is poorly unders. for instance, as many as one in seven lung cancers in women are occurring in nonsmokers, and you

would think that likely has a genetic etiology but that's not well understood. the most common, leukemia, 30% of adult leukemia, no extrinsic factors, so it's a real problem. another big controver is diet. so you'll hear diet spoken about

a lot as a risk factor for cancer, and there are some components of diet that are well known, associated with cancer, like aflo toxin and liver cancer in india. it's been fraught with difficulty studying.

we understand high calories are associated with uterine cancer, obesity, low fiber may have something to do with colon micronutrients, i'm sorry, it's been really, really hard to show that low b vitamins, low selenium, beta carotene, any of these actors really impact

for example, in lung cancer, nutrient-based interventions, there was a gigantic cohort study in finland, 30,000 male smokers, called atbc, and the cohort was randomized between getting vitamin e and beta carotene and many of you probably know the result, the

beta carotene actually increased the rate of lung cancer. at the time beta carotene, animal studies, they were saying i was in the greatest stuff in the world and when the studies came out many of millions of dollars spent on intervention that did harm.

so understanding dietary causes of cancer are really important and really a challe. a lot of controversy about processed versus traditional foods. whether it's the large food groups like proteins or carbohydrates, versus the

micronutrients, that's an area that needs more research. we did a study in italy, where we showed that by consumption, fresh red meat and processed meat associated with increased lung cancer, naturally we adjust for smoking in the study but the truth is that people who consume

a lot of red meat tend to smoke more, and so it's very hard to adjust out those differences. so i'm not personally totally convinced that this is actually a correct finding. well, actually our group mostly there's a lot of gaps on the genetic side as well.

as you know, genome-wide association studies associa common genes with virtually every cancer and most human disease. the problem is that these associations tend to be weak, and when you plug them into risk models, they barely nudge the

risk model. in other words, if i want to decide who am i going to screen in the national lung cancer screening study to have a group at high risk for lung cancer, so that after i screen them they are not a false positive, okay, i'm going to use genetics to

help me. i've asked them already if they are a smoker. i've waited till they've gotten old. maybe they have copd. let's add their genes into that. unfortunately, when you do that study, the genes don't have a

also i have to say that cancer families in complex diseases and what do i mean by a complex disease? i mean a disease that is thought to have both environmental and genetic causes, not a rare genetic syndrome, that mendelizes, or one gene is

likely the etiology. for the complex diseases even with cancer families we have a hard time. currently we have over 50 multiplex cll families, and we've done analysis and exome sequencing and we still can't find a gene that accounts for

cll in these families. it's very difficult. so i'm not going to tell you a lot about genetic studies. i think u know that there are germline changes that are inherited, somatic changes found in the tissue, and there are family studies which is where

you look for rare genes that are determinant and cause you disease, whereas you look in the population, there are common genes that only increase absolute risk a little bit, not so clinically important but can be mechanistically important and there are candidate and agnostic

approaches. fifteen years ago, really before 2007, we did only candidate approaches but after that we've done agnostic genome-wide approaches, and the genome-wide approaches have paid off because we found many new and unsuspects families of disease, families of

genes that are associated with family history is an important risk factor generally, it was one of the reasons we were able to justify doing population studies of common cancers. so when you look at lung cancer and you look at family members, if you have a family member with

lung cancer, your risk of lung cancer increases two-fold for the mother, 1.37 for the father, for any family member 1.57, that's a 57% increased risk after taking into account all the other risk factors. family history by itself is a pretty significant risk factor

overall. as i said, for rare genes you need families. and a lot of tumor suppressed genes were cloned from family this is the kind of diagram you see with a genome-wide association study. we have a long way to go, even

for cancers where we know a lot about the cause. the treatment of lung cancer still is rather difficult. five year survival 15%, targeted therapy really still not having a big impact on mortality. screening for lung cancer, incredibly expensive and

problematic, probles with false positives. so we really have a long way to go. and one of the ways that epidemiology helps traditional epidemiology, i think, i've told you gives people a questionnaire, assesses their

exposure, associates it statistically with disease, based on that assesses factor that you measured as being important. molecular epidemiology takes the next step and characterizes exposure by a biomarker and then tries to determine dose, early

biological effects, altered structure or function, early disease, and disease, essentially it's been described as entering the black box between exposure and disease. so that study i told but in italy, we actually collected blood, and we went into the

operating room, got the tissue, took out the clinical sample, got the pathologists to get us as many as ten pieces of tissue, we got tissue distant from the tumor, we got tissue adjacent to tumor and we got the tumor. and that's all being sequenced now.

and of course we also incorporated all the latest questionnaire effects. this is ten years ago now. one of the latest questionnaires at the time was not do you ask look at doneness, because the idea was the amount of certain classes of carcinogens is

proportional to the time and temperature at which it's cooked so we asked people to look at the doneness of their meat and point to how they like their meat. and, you know, so that was that study i told you about before. it was useful.

so this molecular epidemiology approach has contributed a lot of things, but one we now understand because we're able to analyze the blood samples on people we thought had hpv or in the past we inferred they might have hpv by asking them or by asking their sexual history, now

we now 100% of them have hpv. we know that cutting down on smoking is not really a good idea, because you compensate and you may not compensate in the number of cigarettes you smoke but you compensate in the time that you smoke them in the way that you inhale, the way you

smoke and we have gwas and other biomarkers studies that are good. latest wrinkle on molecular epidemiology is we also look at behaviors and we also look at outcome. so it's also nice to know if all those biomarkers and genes that

we mentioned are related to whether you live or die. i said 15% of lung cancers have longtime survival. well, which 15% and why? wouldn't it be great if we knew ahead of time which ones they were so we could give them more intensive treatment?

so that's the aim of this integrative epidemiology. and we've done some cool studies on this behavioral side. by collecting all this information about the way they smoke, and their mood, and i'll tell you a little bit about that in a second.

the latest -- not the latest, i've been doing this a long time -- is that we now combine our data more readily in consortia. this is a big advance because you can get data, and if your study is not powerful enough you combine with your friends and

you have a much larger study. phenx is an online questionnaire, youd my want to use a questionnaire used by hundreds of other groups and been vetted so if you go there and get questionnaires on anything it's really good. so i already mentioned gaps in

ing of exposure. one of the ways we fill the gaps and i'll conclude and give you a minute for questions, we use emerging technologies to query exposure areas that have been inaccessible so far. so i tried revising these slides, literally i could revise

these slides every two weeks and not quite be up to date. i bought ap thing called misfit a few weeks ago. broke it after a few weeks but it gave a beautiful assessment of how much i sleep. why do i care how much we sleep? we just analyzed nhanes data

which is nationally representative sample, not very good for cancer but it's great for in-depth clinical information, it was clear if you sleep seven to eight hours, if you sleep six hours, your weight is greater, your waist is greater, and that's after

adjusting for exercise, diet, carbohydrates, fat intake, calories, smoking, age and gender. if you sleep six hours or less than six, at six hours you drop off the cliff. and virtually everyone that sleeps less than six hours is

obese. it's amazing. so there's something about sleep that's profoundly important that hasn't been studied because it's difficult to study. you ask people how they sleep and they give you funky answers, but if you have an active

technology measure you get much better, much more in-depth and precise sleep, an extraordinary advance. yeah, here is the misfit thing, i broke my after a week but it's an amazing technology. physical activity, so you all have phones, and you do the

s-health and get the number of steps you did per day, and that's light years better than saying to someone, do you exercise a little bit or a lot or moderate? you know, that's useless. this is spectacularly more informative.

vital signs, heart rate variability, thought to be related to a number of conditions. can now be assessed very precisely. social factors, in framingham by looking at your friends you can tell a lot about your diet,

about obesity, about a number of this is in its infancy but it's something that will be investigated a lot more. location, geography, has an enormous impact on health. and i don't have time to tell but some studies we're doing in this area, but your zip code,

just like your zip code is what on the political analysts look at to determine how you're going to vote and whether it's worthwhile spending money in your zip code to convince you to vote for someone, it's related to held factors as well. and it's been very much

understudied in traditional epi design. smoking, we do a good job with smoking but there are technologies that can help you gather more information and help you quit. i'm interested in weather and climate, and pollution, the

impact of weather on food, characteristics of the water, the biosphere, amount of sun exposure are all understudied areas that impact our health. and finally, last but not least, circadian variation. circadian variation is immensely important, and yet it's

difficult to study in an epidemiologic context because i can't ask, hey, is your internal clock messed up? i can ask if you slept well the other night but in animal models, if you disrupt circadian variation by taking out the p ineal gland or interrupting the

optic tract or doing something that disrupting your internal time, four things happen to the animals. they gain weight, their mood is disrupted. they get metabolic disturbance like diabetes and they have effects on tumors, particularly

tumor progression and incidentals. studying that in humans is very difficult. we use surrogates like asking are you a shift worker, if you're a shift worker i'm inferring your circadian time keeper is disrupted.

what we're trying to do is use metabolomics to develop a blood test for body time. so we're working on that right now with the sleep lab and i'll let you know how that turns out next year. this is a great place to stop where we still have a few

minutes, and if you all have any questions i would be really happen happy. [applause] yes. mike mike [off mic] >> thank you.

>> [off mic] >> so there has been a lot of probably the most information has come out because of the controversial about whether alcohol is truly causal for breast cancer, well established alcohol is associated with

breast cancer but whether it's causal and precisely how it's causal is not so well known. and so there's a lot of wrestling with what the recommendation is. some people i would contend alcohol may have some health effects at very low levels.

currently the recommendations are, memory serves me, one drink a day per less for women, men get two. it's clearly -- it's clearly a breast cancer risk factor and they have done that mendelian randomization analysis i told you about, it seems to suggest

that it is causal. it's very, very clear that it's associated with liver cancer and esophageal cancer so the data there is unequivocal. why we don't hear more about it, i don't know, probably should. probably because, you know, people like to drink and also

there are a lot of, you know, big companies that promote drinking, so, you know, turn on the tv and half of the ads on sundays are beer ads. >> (off mic) >> no. it's the amount of alcohol. obviously you have to drink

lower volumes, and if it's liquor, but as far as i know data on liquor versus wine versus beer, in spite of some minor differences, the idea that wine can contain some compounds in grapes, resveratrol, that's healthy, you know, that doesn't make that big a difference in

epidemiologic studies, and that the risk turns out to be related to the absolute quantity of alcohol consumed and not a particular source of the yes? >> (off mic). >> okay. the question is, what about data

dredging, and gwas, where you have as many as 300,000 to within imputation 30 million genes. because you're starting with so many genes, to test, you're going to find some associations. automatically. how do you get around that?

and the way statisticians have gotten around that point is by establishing what they call a genome wide threshold, and you don't believe any associations you see unless the p-value is less than that genome-wide threshold. how is that established?

the assumption is made, you do a correction, you take the traditional .05 p-value and divide by the number of tests, and we assume that there's about a million genes that are tested, or snps, more than one snp per gene, you divide and get ten to the minus eight.

for an association to be considered proven, in a gwas study, you have to exceed, i say exceed, it looks higher, but the p-value has to be less than ten to the minus eight. we've seen them to the minus seven and they go to the trash bin.

unless it's less than ten to the minus eight you ain't getting it published. that's how that's get with in gwas. >> what about bacteria in the gut, varying from one part of the world to another? >> okay, so folks are going

crazy now wanting to do microbiome studies of the gut to find associations with cancer, and things like let's sequence h. pylori, associated with the number of g.i. cancers, and let's see, you know, how does the sequence of h. pylori vary, and then there's all these other

hypotheses about a leaky gut, is it leaky due to constituents of the diet, some say gluten, some say that's not true, is that leaky gut associated with cancer? that's an area that has not been well investigated. next time.

stay tuned. >> there's always more to do. >> there is. >> we're having technical difficulties. you have to use two pointers, one for this one, one for this one.

>> okay. >> our next speaker -- >> check, one, two. >> our next speaker is kieron dunleavy, he got his medical degree at dublin medical school, in 1994, subsequently did additional fellowship training

st. james hospital in dublin, came to c r in 2002, and he's currently a staff clinician in the lymphoid malignancies branch, he will talk to us today about lymphoma. kieron? >> thanks for the introduction. i'm kieron dunleavy, i work at

the national cancer institute, and i'm a lymphoma specialist so i'm going to talk about lymphomas over the next 45 minutes or so and then we'll have some time for questions, for the first part of this talk lymphomas broadly, what they are, what the different types

are. and then for the latter part of the talk i'm going to talk about some work that we're doing and others are doing in the field. so if you think about lymphoma in the context of other cancers, not hodgkins lymphomas are the fifth most common cancer in

males, and the sixth most c ommon cancer in females. , how common they are to breast cancer, prostate, actually pretty rare diseases. if you look at the difference between non-hodgkin and hodgkin, most are non-hodgkin, 83%, 17% are hodgkin lymphoma, usually

occurring in younger females, most patients present with a mass in the chest. over this talk i'm going to focus on non-hodgkin's lymphoma because they are more common, the most common talk is what i will talk about over the course of the talk.

in the u.s., there are approximately 75,000 new cases of non-hodgkin lymphoma, diagnosed every year. and the biggest risk factor for the development of non-hodgkin lymphoma is increasing age, so if you look at people over 65, and under of 65, over 65 it's 8

per 100,000, these are diseases of older people for the most part. but they do affect patients of all ages, children, young adults, but most people are over 60 when diagnosed with these diseases. shown here are the incidence

rates of non-hodgkin lymphoma in the u.s. over 25 or 30 years, in men and women the incidence has been increasing as you can see, but the good news is that the death rates from non-hodgkin lymphoma in males and females has been decreasing. i think that's because we have

ed improved therapies, and so you can see over the last 15 years there's been a steady decrease in deaths from non-hodgkin lymphomas. so why do people get lymphoma? one of the risk factors, what predisposes you to it? most people who develop

lymphoma, there is no underlying etiology or genetic predisposition that we know of but there are some things that can put you at a higher risk of developing lymphomas. people who have ulcer immunity have increased risk of

non-hodgkin lymphoma. there's another rare non-hodgkin lymphoma, burkitt lymphoma, increased in people with hiv, an important risk factor for development of non-hodgkin lymphoma, and when a patient presents with soms of lymphoma, it's always really

important to do an hiv test balls it can be a presenting feature of hiv, particularly if they have diffuse b cell or burkitt lymphoma. other conditions can predispose you to lymphoma, for example inherented conditions like wiskott-aldrich, iatrogenic like

history of a transplant, on immunosuppressants, these can increase the risk of developing lymphomas, some associated with ebv, some not. patients with chemotherapy for other reasons, for other tumors, have increased risk of developing imfoam a.

autoimmune diseases associated with increased risk, like rheumatoid arthritis, shjogri lupus, treated with immunomodulatory agents, and acquired immune stimulation associated with lymphoma, we heard about infection, heel

co-impactor lymphoma, and hiv as well. there's not really very good studies showing an association between chemical exposure and development much lymphoma but it's hard to investigate in any meticulous way, but i think it's possible that organic solvents

and other chemicals are associated with increased risk of development of lymphomas. what about infections? ebv is associated, burkitt lymphoma is rare, there's one time endemic, that's a type that people in certain parts of the world get, equatorial africa,

children usually between the ages of 2 and 4 develop this, they present with big masses in their neck and jaw and endemic burkitt lymphoma is associated with ebv in 100% of cases, other types are associated with ebv but not as much as endemic, the other two are sporadic and

burkitt associated with hiv, both subtypes ebv is found in the tumor in 30 to 40% of cases. ebv is developed sometimes after a transplant. hiv is associated with aids-related non-hodgkin lymphoma, there is a disease

that is seen in again particular parts of the world in the caribbean, for example, called adult t cell leukemia lymphoma caused by the htl-1 virus, in asia as well, where that's the case these people have a much higher risk of developing lymphoma.

people who are hiv positive there's primary effusion lymphoma, and that's associated with the human herpes virus, ace, multiple myeloma, hepatitis c, helicobacter and other associates, these are the major ones i listed here. so in terms of making a

diagnosis of lymphoma, really it depends on the type of lymphoma in terms of how people present, but most people present with swelling usually of lymph nodes, we have lymphatic tissue in all organs of the body so people can present with symptoms and signs of infiltration of lymph nodes

or organs, and a proportion of patients who present will lymphoma have b symptoms, these are night sweats, fevers and weight loss over 10%, and in certain types of lymphomas those are associated with a worse outcome if you present with those.

because these diseases can affect any organs, they can affect the brain, the bone, the g.i. tract, so people presenting with lymphoma can have a wide symptom of presentation and symptoms. they are neoplasms of lymphocytes, b or c lymphocytes,

the field is more complicated because more lymphomas are classified and more of these areas are being broken down into subclassifications. so there are probably over 80 types of lymphomas, w.h.o. classification, and lymphomas are classified according to

whether they come from a t cell or a b cell. as i showed you, 85% of them are from b-cells, 15% are from t cells. they are classified according to which part of the lymph node they originate from, so there's an area in the lymph node called

the mantle zone, associated with mantle zone lymphoma. they are classified by appearance of cells and hodgkins has a distinctive appearance. you see cells called reid sternberg, and burkitt lymphoma has a starry sky appearance, lots of small lymphocytes,

it's concentrated, an easy diagnosis to make. i talked about this, really over the last ten years we've made significant progress and better understanding the genetic makeup of lymphoma, which are the driver mutations, that's

becoming very exciting now because we now have a lot of small molecule inhibitors targeting specific subtypes of lymphomas, targeting specific pathways that are particularly bad and driving lymphomas to survive. clinical features are also

at the end we'll talk about a pretty rare type of non-hodgkin limb foama primary media steinal b cell that develops in the media steinum, patients have a mediastinal mass, related to hodgkin lymphoma, 80% also present with the mediasteinal, important with the diagnosis of

lymphoma is being made. if i questions so far? what about the breakdown of non-hodgkin lymphoma? as i said, most of these, the most common type is diffuse large b cell lymphoma followed by follicular, the other types are pretty rare, 4 or 5% are

mantle cell lymphoma, more common in asia, in the western world they are rare so you almost always see b cell the two major thoughts are diffuse large b cell and follicular, particularly focusing on b cell during this talk.

and shown here is is a lymph node and you can see the various parts of the lymph node, follicles and follicular lymphomas derived from follicles, there's the mantle zone, lymphoma from there, there's a marginal zone. so different lymphomas that

arise in lymphocytes arise in different parts of the lymph node, and as i said 85% of non-hodgkin lymphoma are b cell lymphoma deriving from different stages of b cell differentiation, so as you can see here this shows b cell differentiation, through the

lymph node, there's a pro b cell, pre-b cell, naive b cell and the germinal center. most lymphoma can be broken down into those derived from the germinal center and those that are called activated b cell lymphomas, and so depending on where the lymphoma originates

from, it's different biologically and clinically usually. this is a busy slide but shows you when a patient presents with lymph nodes and the pathologist looks at the tissue under the microscope and appears like a lymphoma, the way they

subclassify it is typically by doing histochemistry, and this has really changed in the last ten or fifteen years, so for all of these different times of lymphomas there are certain proteins that can be looked for on the cell surface. for example, all about.

cell lymphomas express cd20 and when a person presents with a tentative diagnosis of lymphoma, the pathologist will usually check for cd20, cd10, check t cell markers and as you can see all these have a different pattern of positivity and 20 years ago they had to make

the diagnosis without immunohistochemistries, we can tell which disease is going on. for the pathologist, if it looks like a particular type of lymphoma then they can do certain cyto genetic tests. going back to burkitt lymphoma close to 100% of tumors have got

an a-14 or myc translocation, a lot of lymphomas particularly follicular have a 14-18 translocation, and there's another that can be divided into cases that are alec positive or negative, a protein expressed on the cell and there are immunohistochemical methods to

detect this but these cases have a two-five translocation. when a pathologist is looking at tissue there's no set paradigm about which things should be checked but immunohistochemistry is done first, depending on the suspicious of what type of lymphoma other times of

confirmatory types are done, oncogenes. after the diagnosis of lymphoma is made, it's really important to say that it's very important to get the diagnosis right, there are 80 different types, a lot of infectious processes can mimic lymphoma, so it's critical

but when the lymph node tissue or excisional tissue is looked at, they understand the fitfalls. cases appear like lymphoma but end up being infection, it's not common but it happens enough it's critical to have an expert

hematopathologist sign out the case. after diagnosis they sve an exam with attention to the peripheral node and abdomen, a complete blood count, chemistries checked. there's a test called the lactate dehydrogenase, that's a

marker of tumor blood, something we always check when we newly diagnose patients with lymphoma. everybody should have an hiv test done with diagnosis, screening for hepatitis b and hepatitis c, and the reason is that hepatitis can predispose you to lymphomas but some

therapies can reactivate hepatitis so you need to know if somebody has a history of hepatitis before you start treatment and if they do you need to watch them very closely, they need to have hepatitis viral load checked on every cycle of treatment, and if

there's evidence their liver function is deteriorating you need to pay attention to it quickly because it can be very fatal. patients get imaging done, we do ct scans of the chest, abdomen, pelvis and neck if that's involved.

we are moving away from using cts. there's still gold standards, we ute pet scans, but there are other technologies that are in development and that are not associated with the degree of radiation that ct scans are, so this area is evolving, i think.

it's not uncommon to have involvement of bone marrow when you have lymphoma, 15 or 20%, all should have a bone marrow biopsy. when you're deciding treatment for patients, if they have involvement of bone marrow, it will sometimes prompt you to for

example give treatment directly into the central nervous system. patients with involvement with bone marrow have a higher risk of progressing in the cns, so the cns should be checked and you should at least considr giving intrathecal treatment but that's controversial, i won't

talk about that today. and then we use a stages system, ann arbor staging system divided into four stages. stage one being localized stage two being two or more lymph nodes on the same side of diaphragm, stage three on both sides of the diaphragm, stage

four disseminated disease. this is important, so when we see a patient with aggressive lymphoma, there's the international prognostic index that looks at five characteristics, the age of the patient, the stage of the lymphoma, looks at the ldh

level, something called ecog performance status, and also looks at the number of extranodal sites, several years ago in a very large study these five factors predicted poorer outcomes so patients who had a high ipi score, a score of four or five, had a worse outcome

when they got standard treatment compared to patients with a lower score. i would say that as we get better at understanding the molecular biology of these diseases, these older risk paradigms are not as relevant anymore but there's still

predictive but now i'll talk about in the next few slides, we have a lot of other things that we can look at that are also very predictive of outcomes of patients. so diffuse large b cell is the most common type.

look at the morphology, you see large cells, they are twice the sighs of normal lymphocytes. vesicular nuclei, prominent nucleoli, and t cell rich. so we've made a lot of progress in understanding molecular biology of lymphomas, and i'll talk particularly about the

diffuse t cell and b cell. up until ten or twelve years ago, pathologists were able to say yes, sir this is diffuse large cell b cell, t cell lymphoa, particular markers, cd20 positive, cd10 positive but not able to say anything beyond that.

a group in the u.s. which was led by the lab at nci, lymphoma profiling project, it's a big consortium, of people around the u.s. and canada principally, and they looked at cases of diffuse large b cell lymphoma and were able to break it down into three different diseases, so they

looked at gene-expression profiling, in these cases. they looked at the differential expression of tens of thousands of genes, and when they did that they found that this was not one disease but actually three -- at least three different diseases. one was called deactivated abc

type, one was germinal, gcb, and medialstinal from different differentiation, so gcb time from a germinal, the abc type from a plasmalast and pnbl from a thymic b cell. pnbl lymphoma shares a third of its genes with hodgkin lymphoma, more in common with that.

these three subtypes are characterized by different mechanism of oncogenic of oncogenic of activation. you see the n f kappab pathway,and you see n f kappab, this allowed us to identify some important targets

within these subtypes of lymphoma we can target and hopefully improve outcome for patients with these diseases. where as before and still now we use chemotherapy, and we've made a lot of progress with that but the identification of distinct oncogenic mechanisms has allowed

us to develop novel therapies that are really targeted and can be used within a molecular subtype. so what about the treatment of lymphomas and again i'll talk about diffuse large cell b cell the main treatment for patients with this disease was developed

back in the 1970s, and it is a chemotherapy regimen called chop. p is prednisone. this treatment was developed in the 1970s, and since that time people all over the world have been trying to improve upon the results that they got with chop

back in the 1970s, and they have tried a number of different things. they have tried adding more drugs. they have tried intensifying doing bone marrow transplants. all of these different kinds of and so far, over the last 30

years, the only significant advancement that we've seen has been the addition of rituximab, a monoclonal antibody against cd20, so in france a group there did a randomized study in patients over the age of 60 and hypothesized adding rituximab would increase the cure rate so

patients either got chop or they got chop with rituximab. this is a long-term follow-up. there's a significant survival advantage for patients who got r-chop versus those who got this is long follow-up, these patients are older patients. you can see 36% of them are

cured long term with rituximab and chop. but if you look at patients of all ages with diffuse large cell b cell lymphoma shown here is overall survival and progression-free survival for patients of all ages, plus patients who are low risk by the

ipi that i talked about, they do pretty well, but particularly patients who are intermediate, a significant proportion are not cured. and today r-chop is still the standard treatment, still what most people would give you if you walked into an oncologists

office but there are other strategies in development and this might change soon. this is important because, you know, going back to the molecular biology, that has really allowed us to advance our thinking of what we should be doing to improve the cure rate

of these patients. i think one really important thing that the gcb abc showed, can activation of n f kappab, they are in blue here, you can see overall survival and progression-free survival, one study done, now several studies have been done, for reservers,

it's important to identify this group of people because they have a poor outcome with standard treatment. we need to look at how to do as i said, these abc lymphomas, they have high expression of nf-kappab target genes, 15 nf-kappab target genes shown

here and these are cases of abc on the left and cases of gcb on the right, red shows high expression, green is low expression but you can see these nf-kappab target genes are highly expressed, and the abc subtype, and not in the gcb so a number of years ago, we did

a study at nih where we were really interested in developing new treatments for this abc subtype and we hypothesized that if we develop a target that was specific for nf-kappab, for something in the nf-kappab pathway, then we may see that patients with abc did better

than those with gcb, relapsed we did a study of 60 patients. we used bevacizumab to the gcb and abc patients and found the abc patients had a much better outcome than the gcb patients, a small proof of principle study suggesting that you could preferentially target a subtype.

the field has got more sophisticated, and in abc lymphomas we now have a much better understanding of the various mutations that drive nf-kappab. about 10% of these lymphomas, i'm sorry this pointer isn't

working. 10% have a card 11 mutation, not understood until recently but recent work shows chronic active b cell receptor signaling is important in driving nf-kappa b in these abc lymphomas, and 20% of tumors have got mutations in the b cell receptor, most of

these are cd79b, some are cd79a, driving nf-kappa b, the 88 pathway, 35% of abc lymphomas have got mutations of myd88ase. where chronic signaling is driving nf-kappa b there's a critical step, btk. an inhibitor has made the most

critical in mantle cell, investigated in different lymphomas, but with respect to dlbcl we hypothesize using ibrutianib mild disrupt limb foama where there areings in theb cell receptor and myd88, that shows you that it binds to the bkk active site.

this is work done by lauer staudt. abc and b cell lines were treated. abc cell lines were sensitive. interestingly as you would expect, those abc cell lines that had card 11 mutations were not sensitive to it.

you could expect this because btk is upstream from card 11. based on this rationale, we perform a study in 70 patients who relapsed, they'd bcd and abc, they got single agent ibrutinib until they had progressive disease, what the study showed was that 15 of 39

patients with abc had a response, and a number had a complete response and one patient with gcb had a response. and shown here are the survival curves, if you look at overall survival and pre-survival they were significantly better in the abc patients.

i mean, these curves are not impressive but you have to remember does this patient population in this particular study heavily pre-treated, bone marrow transplants, but the study showed the abc type did better with ibrutinib. this was one patient after the

study who had three previous treatments, you can see this patient had a lot of disease in her abdomen, and after three weeks of ibrutinib you can see a brig improvement in her disease. one very exciting part of the study that was done was we went back to look at mutational

e.status, and correlate that with and patients who had mutants, card 11, for example, they didn't respond to ibrutinib. others are high response rate to it. so this is a very large study which has just recently finished accrual.

and 800 patients have been recruited to the study, the function of the study is in newly diagnosed patients to compare r-chop to standard with r-chop plus ibrutinib, all have the abc subtypes, the results will be available in 2017 or 2018, but if there's a

significant survival advantage in patients who receive ibrutinib this will become the standard for this subtype of so just going to move a little bit laterally, and there's a lymphoma we have a lot of interest here, primary central nervous system lymphoma, so it's

diffuse, almost always of the abc subtype, pretty rare but as you can see over 90% of cases of primary cns are diffuse large cell b cell lymphoma, resembling the abc subtype, the nf-kappa b is activeically active, card11 is mutated, and activating mutations of myd88 are present,

cd79 are common on the left are percentage of cases with myd88, if you look at systemic activated b cell lymphoma under 30% of cases have mutation of myd88. lymphoma called cutaneous leg type lymphoma, 60% have myd88 mutations, primary test

primary testicular is 88%. they are different from the other activated b cell lymphomas, and they also have a very -- they are a -- a high proportion have a high mutation of cd79b, we thought it interesting to test ibrutinib in

this lymphoma, and this is showing you the overlap of myd88 and cd79b in primary cns lymphoma and primary testicular limb ma. we thought it would be interesting to see if ibrutinib is effective. we didn't know when we started

if it crossed the blood-brain barrier, if it would work better in brain lymphoma than in systemic abc lymphoma. i've been talking to you about systemic aggressive lymphoma. if you look at primary treatment of pcnsl it differs from systemic, and the main reason

for that is that the drugs that we use like chop which i mentioned, they just don't cross the blood-brain barrier very prednisone does a little bit but the other drugs don't. for the past 20 or 30 years people have used methotrexate-based regimens, a

chemotherapy drug, it's effective but it's not very effective in systemic aggressive lymphomas, it's been tested in many trials but it's not curative enough that it becomes standard treatment but because it crosses into the blood-brain barrier it's been used as a

standard in primary cns lymphomas. but if you look at the outcome for patients with this disease in general, it's a lot inferior to systemic lymphoma. one thing about primary cns lymphoma, you see a lot of late relapses so we don't typically

see that in systemic lymphomas. but in primary cns lymphoma even if people are doing well as four and five years our field is realizing now a lot of people relapse between five and ten years, for example. so i think what i'm saying is it's very, very difficult to

eradicate this disease, and we can use radiation treatment and that can be very helpful but that is really bad side effects, and it can cause cognitive problems, it can cause vascular problem and these patients who have had radiation treatment are at significantly increased risk

of developing stroke and things like that. there's a real need to develop new treatment for this disease. we basically use the systemic treatment, the standard systemic treatment for large cell lymphoma is a platform and modified it, and we replaced

some of the drugs or used drugs in same class that crossed the blood blood-brain barrier and added it's seen in patients who have hiv but in the study of hiv primary cns lymphoma is almost always evb positive, in patients that don't have hiv it's almost

always ebv negative. we confined our study to patients who were ebv negative, who did not have ebv in their tumor. we wanted to look at the effectiveness of ibrutinib in this disease, also test this new regimen that basically modified

drugs we previously used and in a way that they crossed blood-brain barrier, we tried to correlate outcome. so what did we find? first of all, we did pk studies in all the first group of patients and found ibrutinib did cross the blood-brain barrier,

so patients initially got ibrutinib alone and for the first day had pk levels every two or three hours, shown are the results of pk levels. you can see that red is ibrutinib, the two curves on the top in the first patient we treated, the second patient we

treated, those are both serum levels of ibrutinib, metabolite, two bottom curves are csv levels, ibrutinib gets in for a significant amount of time, and these are the first four patients, we now have twelve patients on the study, and this is looking at ibrutinib

pharmacokinetics. what was interesting was that when we compared the auc and cfs to the plasma of ibrutinib, very little was getting into the cns but in the periphery or serum, ibrutinib is highly plasma protein bound, 97% is plasma protein bound, not the case in

csf. so if you actually correct that, a significant amount of it gets in so in these four people shown here it varied from 21.4% to 100%. this was the very first patient we treated, 61-year-old female, back in 2010, slurred speech,

aphagia, gait instability, had an mri for brain, left cerebellar lesion, they got methotrexate, got the whole brain radiation treatment with another drug and had a response which lasted for three years and then in the middle of 2014 she re-presented with symptoms and

was found to have a left temporal lobe lesion, also in her body she had a left perirenal nodule, recurrent primary cns lymphoma. so she went on her study and got 14 days of ibrutinib, it is given for 14 days and then the chemotherapy drugs are started.

the reason is to see is this agent effective by itself and we do these pks. this was response after 14 days, you can see this lesion in her left temporal lobe has shrunken down with ibrutinib alone. another patient we treated, 65-year-old female, she

presented with seizures, ct showed a right parietal lobe lesion in the brain, primary cns lymphoma, treated with hydros methotrexate and rituximab, good result but relapsed with fourth down in the corpus collosum, you can see the large corpus collosum mass, you can see

improvement in the stars. lesion. this is the outcome of the first six patients that we treated on our study. all but one had a good response to ibrutinib, most that finished the teddi-r treatment also had good response.

ibrutinib gets into the csf and achieves meansful concentration, this treatment is different to the standards used in the past, not so effective. appears to be promising. the plan is to extend to a multi-ry centric study. mediastinal b cell lymphomas.

this was the third subtype i showed at the beginning, if you look at its genetic makeup it has more in common with classic hodgkin lymphoma, they share a third of genes, most patients who get hodgkins lymphoma and media stinal are females in their 20's and 30s, almost all

present with mediastinal mass and often have a cough which started to progress, sometimes the symptoms, night sweats, weight loss, and it's interesting that we now think of mediastinla lymphomas on a spectrum, one end hodgkins, and at the other end you have

primary mediastinal b cell, hodgkin is cd15 and cd30 positive, pnbl is cd20 positive, in between already lymphomas that don't fit into hodgkin or primary mediasstinl, the gray cell lymphomas have histologic and immune oh phenotypic features in between the parent

entities and are transitional, and they are very interesting group of diseases. so how is this disease treated? well, the optimal therapeutic approach is controversial, it's a pretty rare type. there's a real lack of prospective studies and for the

most part it's been treated like others. r-chop has been widely used as the standard. also mediastinll radiation has been administered in a high proportion, the cure rate is very low so it's really critical in this disease to make sure you

give patients the very best treatment at the beginning because if the disease comes back it's difficult to cure it. and the reason that radiation is such a mainstay of treatment is that early studies done 20, 25 years ago suggested that radiation was necessary.

if you look back at historical intensity, as in using regimens more intensive than chop, in retrospect of analyses associated with a much better you have this impression that this disease should be treated with more intensive approaches than chop.

so what about the results of chop, r-chop and in primary mediastinal? this is a mint trial. they went back and this was a trial of six or seven hundred patients, they had a subset with mediastinal. you can see event-free survival

and overall survival which don't look bad but most of these people choose mediastinal radiation treatment. there's a lot of improvement. this study looked at r-chop in primary mediastinal lymphoma from mass general, 63 patients who got r-chop, most rived

mediastinal radiation. people are higher ipi scores, green and purple, had a particularly poor outcome. patients with disease outside the chest, below the diaphragm, advanced stage, those people had a pretty terrible outcome as you can see here.

the blue curves for progression free and overall survival. so we were interested in developing new treatment for this disease. for patients with early stage disease they have a good outcome with r-chop but most of the time radiation is used, and

mediastinal radiation is the gift that keeps on giving, particularly when given to younger people, as they get older their risk of various malignancies in females especially breast cancer, also ischemic heart disease is significantly increased, 20% of

females in their teens or 20s who get media stinal get breast cancer, a portion develop ischemic heart disease. if this comes back, it's difficult to cure. we tested the dosage of da-epoch-r, let's say dose suggested regimen, the drug

based on how low the white count falls after every cycle of chemotherapy, the idea behind that is to normalize drug exposure in different patients that metabolize differently but for most curative regimen there's no strategy to ensure you dose patients based on

having metabolized drugs, just based on height and weight. this was different and we thought this would be particularly effective in this type of disease because historically dose intensity appears like it is important. so this was our treatment

paradigm. we gave six cycles of this treatment and then after that did a ct and pet scan to establish if people were in remission or not. we treated 51 patients here, and 16 patients were treated at a different center, and we didn't

use radiation treatment and found results were very good, this is event-free survival without radiation and overall survival and you can see over 90% of people did really well without any radiation treatment, hopefully without risk of long-term toxicities you get

with radiation. there are a lot of ongoing studies using the regimen in primary mediastinal lymphoma, a study in germany called the g study4 study where this is being used and they found similar results to what we found in our studies.

patients with gray zone lymphoma don't have as good an outcome, these are our results using the same treatment in this disease, and one of our focuses at the moment is to try to understand why the patients with gray zone lymphoma have the worst outcome and we're looking at the genetic

makeup of these tumors and it aprs that they have a genetic signature that's in between hodgkin lymphoma and primary mediastinal bar cell limb foam a in hodgkin the dendritic signature is important, we see that in the cases of gray zone lymphoma we looked at.

the jak 2 kinase signatures are important. in conclusion, i focus on diffuse, but lymphoma is a huge area with 80 types of diseases, but there are a lot of really interesting ongoing studies that are incorporating novel strategies based on our

understanding of tumor biology, and we've made a lot of progress in the last ten or fifteen. and the results may change the standard management of subgroups defined by biological factors. and i think in the future, molecular insight needs to guide clinical trial designs to test

rational and optimal designs. we're closer to having an ideal situation where we move forward and i'm happy to answer any questions for the last few minutes. thank you. >> (off mic). >> no, i mean it's been used in

a lot of different times of patients at this point, the drug is fda approved now, so it doesn't have a lot of -- it has none of the side effects you see can chemotherapy, so some people get nausea, can cause mild gastrointestinal problems, a small proportion of patients get

bleeding. but for the most part it's very well tolerated and most people don't notice any symptoms at all. >> yeah, there have been lots of cases of resistance. that's being studied. so, you know, that's a very

interesting area with all of these novel agents because, yes, of course resistance happens and we're trying to understand why that is, what other pathways are turned on, is there a certain ses to resistance ormutational profile that not. all of these questions.

and in our field we've been using chemotherapy for a long time but i think the future is combining small molecule inhibitors and targeted agents and a lot of the information we're getting from looking at resistance i think will help guide, you know, our -- help us

to decide which are the most logical combinations of drugs to void resistance and things like >> repeat the question. >> you're saying to what degree can gene therapy be used in these lymphomas where there's a genetic predisposition, is that what you're asking?

>> i mean, i think if the biology of the tumor is nf-kappa b is activated, i don't think the gene therapy approaches are going to be particularly effective, but i mean there are lots of -- you heard of car t cell therapy, genetically modifying t cell and reinfusing

here targeting cd19, doing car t cell therapy, cd19 is expressed on most of these aggressive b cell lymphomas but not specifically getting at tumor biology, yes. >> so, yeah, i mean that's a really good question. it's something that's being

explored. there is one type of lymphoma, it's pretty rare but we've been in a position to study it because we have the most cases in the world. what we found is -- again the numbers are small but it's very interesting.

we have been taking blood from patients, lymphomatoid granulomatosis, related to ebv, but what we found is we've looked at blood before and after they get treatment, and i think they get a treatment called interferon for two years, and when we look at their blood

cells and look at their t cells and look at the cd8 compart and go multi-colored flow cytometry when diagnosed with lymphoma, when they are treated and cured the synescense reverses. it's difficult to do these studies but, yes, it's really

interesting i think and, you know, especially ebv is a very interesting causative factor in a lot of lymphomas, a lot can be done in the context of evb might be helpful in cases that are not even ebv positive. >> yeah, it's not entirely clear, is it like chronic

antigen stimulation. people have tried to study this. there are lots of hypotheses but it's still something that's not completely worked out. all right. thanks.