[0:04]Good afternoon or perhaps good morning or good evening wherever you are. It's a great pleasure to welcome you at this seminar in the series of the Department of Epidemiology of the Harvard School of Public Health. And it is my distinct pleasure to introduce to you today's seminar speaker, Dr. Alpa Patel. Dr. Patel is Senior Vice President for Population Science of the American Cancer Society. She is the PI of two really important uh, Cancer Prevention Studies, two and three of the American Cancer Society long-standing cohorts, um, and uh, has a focus obviously on cancer epidemiology. Dr. Patel earned her MPH in Epidemiology at Emery, and PhD in doctoral degree in Preventive Medicine, the concentration in Epidemiology from the Kex School of Medicine at USC. Dr. Patel, it is wonderful to have you and we very much look forward to your seminar, the Science between the Cancer Prevention Studies. May I have your attention for Dr. Alpa Patel. Thank you so much for having me and it's nice to see so many colleagues that I've worked with over the years uh on the call. So I'm going to just give you a very uh quick whirlwind history of about 70 years of of our cohorts with predominantly focusing on the last handful of years. Um, both in the hopes of just uh educating you on some of the work that we're doing here um through these studies at the American Cancer Society, but also hopefully to spark some potential collaboration in the future um that goes beyond much of the work that we're all already doing. Um, for those of you though who don't know about the American Cancer Society, I did want to just start with briefly highlighting that we have three main what we call mission pillars. So uh evidence generation of course um happens in the discovery or research pillar, but then we also do a lot of work that we consider to be actionable that we then bring through education and programs and services to patients through our patient support pillar. Um, as well as work that has policy implications to support our advocacy arm of the organization known as the ACS Cancer Action Network or ACS Can. So there is a lot of movement um in in the work and how we focus our work internally across these three pillars. Um, right now as we're in the midst of uh, you know, just concerns about overall government funding around uh, cancer and other research, I always like to also highlight that um in the research pillar we are the largest non-governmental, non-profit funder of cancer research. Um, in the US we have over a half a billion dollars in grants um out right now. Uh, we do focus while we do have some mechanisms for more uh seasoned investigators, most of our work is um in helping to kickstart um early career investigators with post baccalaureate fellowships, post doctoral fellowships as well as early career awards.
[3:32]So I encourage anyone who um is early in your career to look to ACS for funding opportunities. Um, I'm not going to go into this, but I just wanted to uh, you know, to highlight that we are also the only um not for profit who has both an internal research program as well as of course that extramural funding arm. And so um I sit in the population science program um and oversee that, which is where our cohorts sit. So to jump in, um the American Cancer Society really um I'll draw your line to this volunteers. Um, we have been able to in a very cost effective way um train and engage our community volunteers across the country to build really large cohort studies. Um, obviously this is a Department of Epidemiology so I don't need to go over what a cohort study is, but all of our studies do start with individuals who are cancer-free at time of enrollment. Um, and we do have um, you know, pretty substantial numbers of individuals that have been recruited into these studies um totaling over 2.5 million individuals over the last 70 years. Um, when I think and reflect on these studies, what really stands out to me beyond the size of the cohorts is the long-term follow-up that we have. Again because we recruit through our volunteer base, so we're recruiting friends, family, coworkers, neighbors of volunteers of the organization, we have a really high retention rate for active follow-up. Um, and so that has really allowed us to collect rich data, much like uh, the cohorts that are at Harvard, um nurses, health professionals and and others. Um, it's really allowed us to collect longitudinally updated data over time. Um, and from a from a diversity perspective, I think the geographic diversity is really what stands out to me um in comparison to other cohorts as well as the setting where cancers are diagnosed. So just over half of our cancers are actually diagnosed in true community hospital settings. So we have the opportunity to reflect on factors associated with cancer risk and outcomes in in very much that community environment. So I'm going to very quickly just highlight one or two things from each of these studies to give you a flavor of some of the impact that we've had. And I'll start with the Hammond and Horn study, which only ran for three years.
[6:14]At this time about half of adult men in the United States smoked cigarettes, um, and there was this rise in lung cancer um death rates that had been observed and potentially attributed through smaller case control studies to smoking. Um, this is what uh the these are Doctors Hammond and Horn collecting uh the data by paper surveys. So again I mentioned that, you know, in the early part of the 1900s with the advent of machine rolled cigarettes, there was a much cheaper and faster production of cigarettes, which led to a higher uptake of uh per capita cigarette consumption and then of course this rise in lung cancer death rates. So the British doctor study um in the UK along with the Hammond and Horn study, um, published around the same time to show that there was this excess death um in observed lung cancer deaths as well as coronary uh deaths and total deaths um from among individuals who smoked at least one pack a day. Um, and so this did not require much follow-up, uh what uh is really interesting is that um, Dr. Hoover and Dr. Hammond actually both smoked cigarettes and then switched to pipes, which of course, uh, you know, shows how the evolution of tobacco related research has continued at the American Cancer Society and elsewhere. Um, but to to then show the impact again going back to, you know, the purpose of these studies and why the American Cancer Society has invested in them for as long as they have, is that that decrease in per capita consumption of cigarettes as well as then the subsequent decline in lung cancer death rates were not just through the generation of the evidence alone. But there was also a lot of work to introduce advertising ban, bans and then um excises taxes in uh related to tobacco in all states and then continuing to do work around both both first hand as well as second-hand smoking exposure, the introduction of smoke-free laws in the early 1990s.
[8:42]And so this is where we work very closely with our patient support and advocacy arms of the organization to be able to influence um actual change at the population level. Going forward now to CPS one, what became evident through the the building and uh follow-up of the Hammond and Horn study was that um we had the opportunity to leverage large numbers of volunteers to recruit really large study populations. So we went from 188,000 men to just about one million um individuals across 25 states. Um, we did um include women in this um and enrollment criteria were simple. It was anyone who was over the age of 30 um who had not been diagnosed with cancer was invited to uh provide um information through uh self-administered surveys. And here we expanded on topics beyond smoking. So again uh for again those that are younger in the room who don't know what this looked like, these are the punch cards that were used for data processing and the boxes and boxes of um surveys that were collected from these individuals. Um what was also really interesting is that this is all before the introduction of automated linkage to the national death index. So our volunteers not only recruited these 100 uh sorry one million participants, but they then every six months sent in status updates on those individuals that they recruited to say this person is still alive. If someone was deceased, they went and actually collected the death certificate um to be able to abstract cause of death. So there was really this longitudinal relationship not just with study participants, but with the volunteers who recruited them as well. And what the the study that I wanted to highlight from CPS one was really the first work to document um that increasing body uh weight or composition it was associated with higher risk of overall mortality. So um and this was true for both all cause of cancer as well as cancer deaths overall. So again, really important contributions. So again fast forward roughly a generation in 1982, we launched the Cancer Prevention Study two. Um, this study we did stop follow-up for mortality after 40 years in 2022. So we have a complete 40 years of cause specific mortality. Um, these individuals were um across all 50 states as well as DC and Puerto Rico and completed um an extensive uh lifestyle and health history survey at the start of the 40 years. We continued to build on the early work in tobacco and obesity through the mortality cohort here, continuing to look at now again, not just cigarettes, but pipes and cigars and the burden of various tobacco products on overall health and longevity. And of course, um many of you I know are familiar with the work that was done by the late Jean Cal to understand um cause specific mortality from different types of cancer in relation to um excess body fatness as measured by BMI and really highlighting, you know, the broad range of cancers um that we now of course know are um obesity related. We however, in the early 1990s as we saw that there was increasing survival from cancers, um, there was the need to be able to follow for both cancer incidence as well as mortality. Um, mortality was no longer sufficient. And so, um, a large subset of individuals in that 1.2 million cohort were invited if they lived in one of 21 states with at that time high quality population based cancer registries were invited to join what we call the CPS2 Nutrition Cohort. This subpopulation of about 190,000 individuals is followed for both cancer incidence as well as mortality. Um, and beginning in 1997 and every two years thereafter, we did um administer an updated survey to update uh lifestyle and other factors that may have changed. So to give you an idea of just some of the things that we've collected over time out of the 100 after some exclusions 184,000, we went back in the late 1990s and invited uh nearly 40,000 individuals who were near a Commission on Cancer Hospital to provide a blood sample. We augmented germline DNA with an additional 70,000 individuals uh to whom we mailed a mouthwash kit to get buckle or cheek DNA, um and then we did um collection of uh tumor tissue for select cancer types, uh retrospectively in the late uh right around 2010. We also because of the repeat follow-up and maintenance of addresses over time, um everyone is geocoded at every known address and we've linked to a variety of environmental data sources. Um we've also used identifiers to link to all parts of Medicare beginning in 1999 when the vast majority of our participants were age 65 and older. So to give you a couple of um highlights of CPS2 Nutrition Cohort, um this was, you know, an important contribution around um the impact of consumption of red and processed meat with the incidence of colorectal cancer. Um, and uh one of the earliest studies that that really highlighted this um in a prospective study design. Um, I actually joined um in the late 1990s here at the American Cancer Society and have focused the majority of my work on physical activity and cancer. Because um a couple of years before joining here, um in the Surgeon General's Report on Physical Activity and Health, there was one sentence um in the report that said that regular physical activity may lower risk of colon cancer, but data were too sparse for really any other cancer. So uh felt like a a nice area to focus my uh career path. And so I spent a number of years looking across a number of different cancer types, but then what we did is we actually came together um with some of the cohorts at Harvard along with a handful of other cohorts to do a robust uh pooled analysis with studies that similarly asked about physical activity.
[16:01]Um, to actually um look at what the relationship was across a number of different individual cancer types. And if we look today at the state of the evidence in the last um US report from the physical activity guidelines committee, there are now seven types of cancer that are considered conclusively or convincingly linked to physical inactivity. Um, so this is a lot of progress, um although of course we still have the challenge of how to increase physical activity levels among those who are insufficiently active. Um, we similarly have seen that there has been this decline in overall physical activity with an increase in in sedentary time. And so we've done some work to look at how the combination of these movement behaviors interact with one another. And I thought this was a really interesting paper that we were able to do in CPS2 Nutrition Cohort, where using iso-temporal substitution modeling, we modeled that if you were highly active as measured by your moderate to vigorous physical activity versus low active. Um, if you replaced 30 minutes of sitting time for either moderate to vigorous physical activity or even light intensity physical activity, we see a we see a reduction in all cause mortality risk. And so this is a really nice message that if you are highly active, the amount of sitting time you do or increase or decrease may not have as much of an effect on um premature mortality. But for those that are moderately or low low active levels, um the replacement of even 30 minutes of sedentary time could be very beneficial. We also again to bring things circle, we're always evaluating the state of the evidence and our guidelines. Um, this is some work that we did to show that um, if you follow our American Cancer Society nutrition and physical activity guidelines, um, the the better it is for you. So again, a nice take home message, you don't have to strive to perfection, but the greater the adherence to the guidelines, so zero to two is people who minimally um may follow some aspect of our nutrition and physical activity guidelines. Seven to eight are individuals who are highly adherent to all aspects of the weight maintenance, physical activity, diet pattern and alcohol guidelines. Um, but you can see this really nice dose response with overall mortality and you know, uh it it more plateaus for uh cancer mortality, but um in both men and women there's, you know, a nice um benefit where even partial or full adherence to the guidelines is important for overall and cancer specific mortality. More recently in the last handful of years with um all of the the repeat data that we've collected as well as um the specimens that we've collected. This is really opened up a lot of opportunities to not only look at survival, but also to look at aspects of survivorship. Um, the goal is to help individuals with cancer not only live longer, but live with a high quality of life um as well. So we did some work again bringing in our Medicare data where we were able to look at um claims around fractures. Um, individuals who are aging but without a cancer history as well as um cancer survivors, to look at whether or not um the risk of fracture, which we know is a problem with aging as a whole, but uh more pronounced among individuals with a history of cancer. Um, can we actually find ways to intervene to lessen some of that risk? And what we actually found is that individuals who engage in um high levels of moderate to vigorous physical activity and strength training actually do have a lower risk of um fractures. And so this this type of, you know, information can be really important in helping um survivors uh work with exercise professionals as well as we saw, you know, very strong association with smoking. Um that smoking cessation programs to be able to um modify some of that future fracture risk. Um, using the germ line uh DNA that we've collected from um collectively between blood and buckle cells 110,000 individuals. Um, we also want to be able to look at things like polygenic risk scores. And so here what we did is among breast cancer survivors, um for whom we had the polygenic risk score, we looked at um tertiles of polygenic risk. And this is polygenic risk of obesity, not of breast cancer. This is not the polygenic risk score for breast cancer because these are already individuals who have been diagnosed with breast cancer. Um, that we see that um if you have a genetic genetically higher uh propensity for obesity that you are at a higher risk of mortality after a breast cancer diagnosis. Um, what we absolutely don't want to do is though say, well you're genetically, you know, there's doom and gloom. Um, so what we then did is we looked at based on your underlying genetic risk. So to orient you, um those with uh in the green line are those with the highest um genetic risk uh for obesity. If you walk, you may have to walk a little bit more per week um closer to three hours to mitigate that genetic risk than if you are at low genetic risk. So um again, a really actionable message that how do we take into account your genetic predisposition to survival, but then fold in opportunities to address that genetic predisposition through modifiable risk factors. Um, so you're not doomed by your genetics and we can actually mitigate that risk. It just may be that you have to engage in a slightly higher level of that lifestyle behavior like walking. So we've also used beyond survivorship about one in every 10 new primary cancer diagnosis is a second cancer among individuals with a cancer history. So it is really important to understand for individuals that have had a cancer diagnosis, how can we help reduce their risk of a second cancer diagnosis because we know that they are at higher risk for a subsequent diagnosis. So among our nearly 30,000 um survivors of a first cancer, um we have a large number with Medicare data, a large number with genotyping data, but we also have the overlap of those. We have about 8,000 individuals um for whom we have both Medicare as well as genetic information. And so in this group what um we are what we've been looking at and this is a paper that um is under review by um Clara Bodellon in my group, um is we're looking at um a pan cancer polygenic risk score and risk of a second cancer. So we know that polygenic risk scores put you at higher risk uh for certain types of cancer for that first cancer, but what we actually see is that that genetic risk carries forward even into your second cancer as well. Um and so it's important to think about how we need to incorporate genetics um into ongoing surveillance of individuals with a history of cancer. We've looked though at other factors as well. So we've looked at um the association of excess body weight, which we know is associated with at least 13 types of cancer. What does that mean for your risk of developing a second cancer if you've already had a cancer? So these are not just individuals that have had a first cancer that was obesity related, but have had any cancer and based on their BMI, we see that their risk of a second cancer goes up with higher BMI. Um and even more so if that second cancer is obesity related. And we've similarly done this with smoking as well. So we've looked at individuals based on their smoking history, um whether they are currently smoking, whether they are former and have quit and by how long ago they quit. And what we actually see is that um even for those individuals that quit 30 plus years ago, there is a modest elevation of risk of developing a second cancer. And so again, how do we think about perhaps more active surveillance um of these individuals? So switching now to the last of the the cohorts is the Cancer Prevention Study three, which I've had the privilege to lead um, you know, from from inception. Um and this study is just over 300,000 individuals between the ages of 30 and 65 at enrollment. It is smaller than CPS one and CPS two because we made the investment to collect and bank a blood sample on everyone at the time of enrollment so that we were well positioned to be able to incorporate genetic and other risk that can be measured through these blood samples. Um of course, the, you know, it's important to think about these generational cohorts from the perspective of new and emerging um exposures. So if we look at CPS two, um less than 20% of CPS two participants um had a BMI um that would be considered in the WHO cutpoints of overweight or obese categories at age 18. About 50% of CPS3 participants are in those categories beginning at age 18, so we don't completely understand the lifetime health effects of having obesity over the life course and what that may mean. Um so these types of more contemporary studies will allow us to examine these. So CPS3 much like CPS two, uh we follow the entire cohort with uh surveys, uh long surveys electronically or on paper. There are still individuals who don't trust computer-based surveys and want a paper survey um every three years instead of every two to update um those lifestyle and health history uh data points. Um but we've also done a number of things to embed um various substudies into um the cohort. So um, we have, you know, for instance like CPS two, we have geocoded all of our addresses and have linked to various built environment as well as other environmental um pollutant databases. We are collecting tumor tissue, we're working to expand our tumor tissue collection to include and endometrial and lung cancers as well. Um we completed accelerometry um research grade accelerometer collection on um just over 21,000 individuals. Um we have onboarded about 70,000 participants into a participant portal where we have they have consented to be approached three to four times per year for invitation to provide more survey-based and other data. So things like a 30 a video-based 30 second sit stand chair test and other types of things that we can now administer remotely um to participants. Uh we use the portal for that. Um we completed collection of um 13,000 participants with paired um saliva and stool samples for the oral and gut microbiome. Um we uh worked um actually with uh Andy Chan and others uh to push out the uh symptom tracking app for COVID. Um and uh we have also embedded our first behavioral intervention in 400 survivors of physical inactivity related cancers to work on a one-year-long um web-based intervention to increase physical activity among cancer survivors. And we just wrapped up collection of repeat blood samples on um a large subset of participants. Um so these data have allowed us to start getting into again some new and emerging areas. Things like um cannabis use and their relationship with uh cigarettes. Um this is a lot of work that is influenced by our advocacy colleagues um around what legislation should look like um for um the the legalization of cannabis, but in this study we looked at whether or not um smoking marijuana was actually um influenced the your ability to quit um combustible cigarettes and found that there was no association between um the quit rates of either. We've also been um working in the space of multi-cancer early detection in in a couple of different ways. Um and and I want to just share uh a couple of these examples with you um because I think that this is a a newer um and far more novel use of these types of um prospective um data where we've collected blood samples on individuals prior to a cancer diagnosis but know that exact interval between blood collection and diagnosis. So this is some work I presented um that's under review right now at ASCO. Um a couple of year about a year and a half ago where we actually took samples from CPS3 where uh we know, you know, in this blood draw window when the blood was drawn and when a clinical or conventional diagnosis of cancer occurred. We know exactly to the day, how many days um included. And so the question becomes could we detect cancer earlier than that uh conventional diagnosis? Um and what we actually found even though we did not collect these samples in an optimal way, we used EDTA tubes versus strep tubes, which, you know, affected the overall the absolute detection rate. We actually saw that in particular for cancers that are um more aggressive or eventually fatal, um we were uh we were capturing those and able to detect those cancers. Um and we were less likely to uh cap or identify those cancers that um were not um as aggressive. Um and this becomes really important.
[32:06]It's not only that we were able to detect that cancer signal, yes, no, but um through these blood samples, uh a very high level of accuracy in terms of the cancer signal of origin. So for those that uh were within that first year about 95% accuracy in terms of where those cancers were originating. Um and so this is some work that we're continuing to do in large part uh through industry collaborations. We're also though working to understand um in this era of, you know, it's it's probably not a matter of if, but a matter of when some of these multi-cancer tests become uh get FDA approval and become more readily available. They're already available direct to consumer with a physician order. Um but who should who may benefit from these tests and how do we think about risk of all cancers? When we think about individual cancer screening tests, we base our screening recommendations on who and how risk is impacted for breast cancer specifically, for lung cancer specifically, colorectal cancer specifically. But when we're thinking about multi-cancer detection, um these are who is at risk of any type of cancer? And so is age alone adequate? And so this was some work that we did to actually understand five-year absolute risk. And so the solid black line is if we took just age alone, what the age curve uh from seer looks like for males and females. But then if we add in smoking history, body mass index and family history of any type of cancer, we actually can see that there's discriminatory power in absolute cancer risk. So if we look at for instance an individual who has a BMI greater than 30 and is currently smoking at age 45, they have a higher absolute risk than individuals who have none of those risk factors, a full five years later. So how should we be thinking about the inclusion of other easy to measure risk factors in um screening in enhanced screening recommendations? I want to just end then by talking about our newest initiative, which um after some pilot work in late 2023, we launched last year. Um, and that is a new study, our first uh focusing on a specific population, but it's called Voices of Black Women. Um, historically, while we got incrementally better at racial ethnic diversity in our cohorts, we really looked at um where there were significant gaps in representation in these types of large population studies. And there are other uh large initiatives underway to address other um underrepresented minority populations. But um so that's that was really why we are starting with Voices of Black Women. But our goal here is to really understand the combination of factors that contribute to cancer risk and outcomes in black women to facilitate more uh focused opportunities for intervention um in this population. So we'll go beyond those factors that are assessed in most um population studies like comorbidities and um lifestyle and environmental behaviors, but we'll really dig into social determinants of health and aspects of individual all the way to systemic racism and cultural and other factors as well.
