David Gozal, MD., Herbert T. Abelson Professor and Chairman of Pediatrics, and Physician-in-Chief, discusses the possible connection between OSA and Cancer.
SPEAKER 1: You might wonder what we do in pediatrics to start talking about sleep and cancer. But this was a little bit of a serendipitous encounter. I was visiting my collaborators in Barcelona many years ago. And we were wondering as part of-- and you will see some of the data that has really emerged as a very important as a preliminary set of data. But we were wondering whether this concept of intermittent hypoxia that occurs during sleep apnea could potentially contribute to cancer biology. And at the same time that we were having those discussions, multiple investigators in the field of cancer were really starting to unravel some of the mechanisms by which the tumor would recognize or be subjected to intermittent events of hypoxia because of lack of profusion, accelerated growth, so that very aggressive tumors seem to have even more aggressive behavior. And this seemed quite intriguing. So with that, we decided that it would be worthwhile to start exploring this concept. So what I'd like to do is to talk to you about human studies, introduce a little bit of what is linked sleep, because I think that it goes beyond, and I hope to convince you that it goes beyond intermittent hypoxia alone, that we're really talking about sleep and sleep disorders, and that we will show you some of the human studies that really anteceded some of the work that was done. And then I will present to you two series of data-- one on the animal models and some of it done in Barcelona-- and I'll introduce who did it because he's in the audience-- and also work that was done here, and then ultimately close with some potential conclusions and directions. So there's a study that was started many years go, over 20 years ago, in Wisconsin in Madison, the University of Wisconsin at Madison. It's been a longitudinal cohort. It's called the Wisconsin cohort. And that cohort published about 2008 or so published 18 years of experience. And they looked at mortality. And the reason that they were looking at mortality is that they wanted to explore whether sleep apnea as a disease was going to affect the overall mortality rates. And what they found is that, as you can see in this graph, is in a couple of [INAUDIBLE] is that as the severity of sleep apnea illustrated here by a concept of apnea hypopnea index or AHI that as it was more severe-- in other words, the more severe the sleep apnea-- the more likely that you would die, and your life would be curtailed. The interesting thing is that buried in that paper the vast majority of the cases who died did not die of cardiovascular disease as we suspected, but actually died of cancer. And this went essentially unrecognized. So it's an interesting observation. And as many of you know, cardiovascular disease is one of the consequences or the major consequences of morbidities of sleep apnea. And such, it was somewhat surprising that the reason that you would die was not really cardiovascular but rather cancer. So I just made this quick graph to illustrate a little bit what sleep apnea is for those of you who don't know it. And for those of you who do, I apologize for boring you. But sleep apnea in the context of adults is related to the upper AV obstruction during sleep. During that process, you have cessation of flow, ongoing respiratory effort. So somebody's struggling to open that away. And in that context, you have your oxygen-hemoglobin saturation will be reduced, and ultimately you will arouse. And you will wake up just to open your airway, fall asleep, and, again, repeat this cycle. And severe patients with AHIs of a 30, 40, or 50 as the number would indicate would have 30, 40, or 50 events per hour, which obviously should not be a very healthy thing. You can imagine that if we get a phone call every two minutes or so, the next day you're a tough person to talk to. If, on top of it, your oxygen delivery to your brain and to all the other organs is compromised, there's obviously going to be multiple consequences. And these have been explored extensively, and our lab has been one of those in which symptoms-- such as cardiovascular, metabolic consequences, neural cognitive consequences-- have now been very well recognized and very well illustrated. But if we really think about sleep apnea as a whole and possibly sleep disorders, one of the major consequences is being the activation of inflammatory and reactive oxygen pathways. So you have oxidative stress. You have information. And those could potentially be the key mechanism that leads to many of the end organ morbidities associated with sleep apnea. One of those, if you think about inflammation and oxidative stress, both of these conditions have been associated with cancer as well. And so it would not be a far-fetched supposition that the same mechanisms that are triggered and pathological processes that are triggered by sleep apnea will now be involved possibly in eliciting some degree of tumorigenesis or cancer-related effects. Now, the other element that we need to think about is that obesity, which is one of the major associated features of adult patients with sleep apnea, has been now very well recognized as a contributor to cancer. Now, if you recall very well, is that obesity is a systemic low-grade inflammatory disease associated with oxidative stress. So again, the patient with sleep apnea has some degree of opesity, sometimes very obese, therefore is already increasing with risk. And on top of it now, you have this obstructive apnea, which will further amplify and exacerbate the inflammatory and oxidative process such that you're now doubly or synergistically increasing the cancer risk and the risk for cancer progression. Now, the issue of inflammation and cancer is not new. This has been now a subject of many, many, many research. And we know, for example, that inspections lead to certain low-grade inflammatory. We know, for example, the patients with inflammatory bowel disease are at risk. So we know that inflammation, chronic low-grade inflammatory processes, are underlying a increase risk for cancer. And some of those mechanisms are being just unraveled as we speak. And they affect some of the mediators of inflammation. The NF kappa B induction of a variety of transcription factors that have to do with tumorigenesis and metastatic potential. So another interesting study done by the Japanese-- this was also reproduced here by Dr. [INAUDIBLE] and a very similar cohort of women were at risk for breast cancer. What they found is that the sleep duration-- in other words, some degree-- and this, of course-- these are studies that have not been designed to study sleep or sleep quality but add a little question. Just, how many hours do you sleep? And when they looked at the risk for cancer, in post-menopausal women, what they found was that post-menopausal women had a higher risk if they slept less. Now, slept less doesn't mean that sleep is normal. And sleep more doesn't mean that the sleep is OK. So obviously, there could be a little bit of confusion. But nevertheless, this, again, would suggest that alterations in sleep to a certain extent could be linked in some way to tumor biology as well. And I've alluded to the concept when we were discussing intratumoral hypoxia, and the issue that hypoxia is going to happen tumors, particularly in core areas. You know that many of the tumors that grow fast have necrosis. And the necrosis is possibly because there's not enough supply. There's a variety of activating factors, such as HIF, hypoxia-inducible factors that are turned on. And some of that work has shown that in certain cells, HIF-1 alpha or HIF-2 alpha, have a major role in both the behavior of the tumor in its progression as well as the potential for mometastasis. So all this leads to the perfect storm, the perfect storm of sleep apnea, according to the way I look at it. You have intermittent hypoxia. You have disrupted sleep. If you have those willing to really be associated with an increased process of cancer. So we formulated a hypothesis-- and this is illustrated in the page paper that will come out in the next few weeks or so-- where obstructed sleep apnea associated with intermittent hypoxia with sleep fragmentation would lead to increased tumor malignancy, but through a mechanism that we hypothesized was related to alterations in the immune system. And what I'm going to spend a little bit of time now is to try and show you that indeed this is the case. So the first question that we have asked in the laboratory is, does intermittent hypoxia alter the biological properties of imuran lung solid tumor. We picked a solid tumor that is very well established in the literature. We're not trying to innovate in that context, but really to understand some of the more fundamental alterations that could have with sleep or with intermittent hypoxia in the context of cancer. And so the question was, does it affect the proliferation of that tumor, and does it really affect the malignant properties of the tumor cells themselves? It could be that you may not change the growth of the tumor, but it could affect the way that the tumor penetrates into other areas and becomes more invasive, for example. So I'm going to go back. And Dr. Almendros is here in the audience. You will recognize him because he's going to raise the hand. But he was a postdoctoral fellow in Barcelona, at the University of Barcelona, and Professor Farray, one of our collaborators. He start an outstanding series of very revealing experiments that tested very simple questions. And the first question is, does intermittent hypoxia modify cancer progression? And so what he did-- he took a melanoma model in this case. Melanomas are very rapidly growing tumors, as you all know. He injected them into the flank of the mouse. And what he subjected them was to a model of intermittent hyposia that mimics, in a way, a moderate to severe patient with sleep apnea. And he did it for six hours a day for 15 days. This was obviously a relatively more severe, as I indicated, sleep apnea. And this is a tumor. This is the way the tumor looks when injected into the flank. So you can see that it's quite an interesting large tumor. And when he looked at the progression of the tumor, as shown here, you can see that the tumor that's exposed to intermittent hypoxia grew much faster. And when he took those tumors out and excised them and actually weighted them, the tumor also weighted significantly more, suggesting that intermittent hypoxia will-- and during sleep, during six hours corresponding to the sleep period-- will accelerate tumor proliferation and growth. So he then asked whether the same model, whether you would have an increased tendency for a very metastatic tumor. As you know, melanomas are exceedingly metastatic. Will that really lead to increased number of the metastasis? Melanoma, as in this mouse model, will go to the lung. And so they're easy to recognize because they have the pigment, melanine, such that this was a relatively straightforward model. And therefore very intuitively, what they did was to subject the mice again to this exact same model. And here shown is the pigment containing areas that are tumor cells that have penetrated into the lung. And those are easy to count. He counted those and what they found was that not only there was an increased area of intermittent hypoxia exposed mice. There was an increased areas of metastaci tumor. But that the number of metastasis was increased, significantly increased. And the size of each metastasis seemed to be the same. Suggesting that it was not just the metastasis perceived will not be different, but more importantly, the tendency to metastasize in the context of intermittent hypoxia appears to be enhanced, therefore potentially associated with adverse prognosis. And we'll talk a little bit about that later when I present to you some of the clinical cohorts that are available. Now, illustrating this, what Isaac saw was that the mice that were exposed to intermittent hypoxia and had melanoma were more likely to die than those that were not subjected to intermittent hypoxia. So the summary of the results for this series of experiments, which are really the first step in convincing that there is a biological plausibility to what I was alluding to is that intermittent hypoxia during sleep increases melanoma tumor size and metastasis from the primary subcutaneous tumor. But what about other tumors, such as another tumor that is a [INAUDIBLE] epithelial lung tumor, a TC-1 cell. This is a very well-established model. And we asked another question when Isaac transferred to the University of Chicago. Is this the tumor cells, or is it the host that is actually modifying this problem? Is it really the tumor itself that is changing because of intermittent hypoxia, or is the host playing a role in the context of these observations that I've reported to you? So this graph is a little bit of a busy graph that illustrates-- on the left side here, you can see that the tumor growth pattern of TC-1 tumors in [INAUDIBLE]. These are individual trajectories being weighted and measured until the animals are sacrificed on day 28. And you can clearly see that the mice exposed to intermittent hypoxia have an accelerated growth, as illustrated also here that the overall tumor weight is significantly higher than in mice that are not subjected to intermittent hypoxia. But there was also a very interesting observation. And the interesting observation was that if you looked at the invasion of this tumor locally, there were tumors that were in mice that were exposed to intermittent hypoxia were much more likely to invade the surrounding tissues. And this is shown here. This is a normal process. This is a standard tumor in [INAUDIBLE]. And you can see that the capsule is around. This is a very well-defined capsule. And the tumor cells are separate from this capsule and the surrounding muscle. So it seems to be contained very nicely in a disencapsulated form, which is the typical phenotype of this tumor in mice and described for many years. But when we looked at the mice that were exposed to intermittent hypoxia, there was disruption of the capsule and invasion of tumor cells both of the capsule and to the muscle. And this is shown here as well. And when we counted blindly for how many of these tumors in a total of very large number of mice you can see is about 40 mice in each group. When we counted how many it invaded [INAUDIBLE] mice, there was only nine of 40, while, in the intermittent hypoxia, it was 24 out of 40. So clearly a very significant increase in the local invasion that occurred during intermittent hypoxia. So we asked a second question. Well, can we reproduce this in vitro? Because, after all, the in vivo conditions-- can we take some of these characteristics and go and do that in vitro? And what we did here was to take TC-1 cells and take now macrophages from the mouse that was exposed either to intermittent hypoxia or to [INAUDIBLE]. And we saw is that while macrophages coming from the mouth that was exposed to [INAUDIBLE] increased maybe a little bit. The proliferation of these now completely naive TC-1 cells. When we took macrophages that came from mice that were exposed to intermittent hypoxia, the proliferation was explosive the accelerator. So suggesting that maybe macrophages are involved in this process of accelerated tumorigenesis. And the interesting part was that indeed the tumors that were excised from mice exposed to intermittent hypoxia had the much larger number of these tumor-associated macrophages. Now, this is now a very well-established marker for adverse prognosis in human cancer. If you take cancer, solid tumors and look at the number of tumor-associated macrophages, the larger the number of those cells, the worst the prognosis of that tumor and the more aggressive behavior and adverse prognosis you can see. And this is exactly what we found. And so, again, as you looked at the number of these cells, we saw that indeed the proliferation was accelerating. But what about migration? So the first step of invasion is migration. And so what Isaac did was to really look as to whether tumor-associated macropages-- I'm going to call them by their nickname, TAMs. Whether TAMs of mice exposed to [INAUDIBLE] or to intermittent hypoxia would induce migration of these tumor cells. And lo and behold, what we found is that when you were exposed to intermittent hypoxia, there was enhanced migration of the tumor, OK? And this is in vitro. So these are naive cells. Suggesting that, again, the host is playing a major role, because it's promoting both the proliferation and the migration. And the last thing is, can you really invade a matrix such as the capsule? And so to do that-- this is shown here in these little circles that are very cute and very nice looking. But what they are-- this is a three-dimensional matrix in which we have put tumor cells. And what we have done is exposed them to intermittent hypoxia. And what we see is that when you expose and put them with macrophages, the proliferation goes into three-dimensional space. And so the more they are able to cross the matrix, the more invasive they are. And you can see that the circle is clearly much bigger in intermittent hypoxia, suggesting that indeed these tumor cells, when exposed to these macrophages that FC intermittent hypoxia can now not only start their migration but really cross the physical boundaries and really traverse and then become locally metastatic. So we call this a process of invasion. But we want to test it in a different way, because you can have invasion locally to another issue, such as a capsule. But you can have it in two vessels, right? Which is then how metastasis really get formed. And so we did an interesting study, which is shown here. I'm going just to spend a little time. What we did was to take the tumor cells, have them grow-- excuse me, endothelial cells-- have them grow as a single layer. And then we took the tumor that was exposed to either intermittent hypoxia or to [INAUDIBLE] and measured how these cells disrupted the endothelial barrier. And by measuring the constant of electricity or the resistance to electricity of a constant current, you can actually identify that if you have a decrease in the resistance, that as being penetration into this layer, this endothelial layer. And what you see here is the tumors that were exposed to intermittent hypoxia, particularly when they were accompanied, tumor cells with my macrophages really disrupted the endothelial barrier. So you have all the elements that would suggest intermittent hypoxia is associated with both increased proliferation, particularly through the presence of macrophages or TAMs to that they will accelerate or increase migration. Three, that they will increase invasion and potentially have an increased metastatic potential, because they're able also to cross into the vascular space, OK? The third one that we wanted to test is whether indeed there was a difference between culture of tumor cells in intermittent hypoxia alone, or whether they would differ if we put them with macrophages. And this is a very interesting finding. And I'm going just to focus here for the sake of time. But what Isaac found was when he exposed tumor cells alone to intermittent hypoxia. There was no difference in their growth. You needed to add the macrophages. These are naive macrophages coming from a different cell. So these are not preconditioned. And yet, by exposing these macrophages to intermittent hypoxia, you now change the fetal type of these macrophages. And these macrophages are now eliciting or inducing, sending all sorts of messages that need to be defined yet. But they're sending all sorts of potentially growth factors or other mediators that are eliciting now and accelerated proliferation of the two. So the other element that we found was that as part of this intermittent hypoxia exposure, the macrophages, when exposed to intermittent hypoxia, changed polarity. Now, what do I mean by polarity of macrophages? For those of you who may not leave your daily life thinking about macrophages, macrophages are a very heterogeneous population. They're not as we would like to believe, that they're all the same, but they're not the same. And over time, it has become apparent that, in fact, there are many, many, many, many phenotypes. Now, we're talking today about resident macrophages and bone marrow derived macrophages. We're talking about within those categories those that are more of, what we call, pre-inflammatory type or N-1. They're the typical that produce TMF in relation to the response, or IL-6 or IL-1 or interferon gamma. And there are those that are a little bit not as inflammatory that are actually anti-inflammatory to a certain extent, and we've been called them too. And they produce IL-4 and IL-10. They can be induced with TGF. So they're very different macrophages and they behave very differently. And so tumor-associated macrophages are more likely to be type 2, M-2. And that's why they've been called M-2 light, because they're residing them in the tumor. They may be a little bit different, and they have a little bit different markers. And so here what we did was to look at the changes that occur in this polarity of macrophages. And what we found was that, as shown here, is that when you expose macrophages to intermittent hypoxia, they become more pro-tumor M-2, OK? So if you remember that M-2 is a bad thing for a tumor-- it's a good thing for the tumor, maybe, but not for us-- then to see a transformation of the polarity of macrophages to an M-2, a preferentially M-2, in the spectrum is not a good thing. And so one of our collaborators, Lev Becker, who is also here-- we asked him if he could look much more unbiased, not just by looking at the single marker-- trying to do a mass spectrometry unbiased assessment of the membranes of these macrophages that appear in tumors exposed to either intermittent hypoxia or to [INAUDIBLE]. And what he found using a very sophisticated mass spectrometry approach, was that they were many, many markers that are really associated with the type 2-- the pro-eugenic, pro-lung inflammatory that promotes this drawable proliferation and has all these characteristics that would induce tumor progression. And so we got confirmation both from a hypothesis-driven single marker but also from an unbiased approach. And we are convinced that this really now provides us with opportunities of maybe identifying specific targets that we could regulate and change in the context of tumor growth. So this also provides us with a lot of information that may be very important in the future to identify therapeutic targets in trying to convert macrophages from one type to the other and fight the tumor more effectively. So to summarize the results just on the intermittent hypoxia is that there's no doubt that more aggressive tumor phenotype is observed in the context of our age and is dependent on alterations of the hosting end response, we believe, in general, and more specifically in the macrophage polarity from shape-shifting the paradigm from more of an M-2 to more of an M-2, OK? But what about sleep? You came here to hear also about sleep, and so we needed to look at the other component that I showed you that characterizes sleep apnea, which is fragmented sleep. So would sleep fragmentation do anything? So in order to do sleep fragmentation, it's not so simple. We spent a lot of years developing a model that would have many characteristics. And the characteristics of this model, which was really developed with Dr. Huang and Shelly Jang in the laboratory was to really look at the model that would comply with several criteria. One, it needed to be non-stressful. As you know, the presence of chronic stress can accelerate too. And so if we had something that is stressful, it could be attributed to the chronic stress rather than to the actual perturbation of sleep. Second, the animals should be completely freely moving. They cannot be restrained or not allowed to move the way they want. Three, they need to be kept in their normal social setting. Mice are very social animals. And if you isolate them, then as many of the methodologies today used to disrupt sleep, then these mice are going to be stressed. So social isolation is not something good. So we needed to allow them to keep in their normal environment. And they needed to be able to have access to food and water as much as they wanted and avoid any human contact. Mice have a very sensitive ability to smell and to engage. And if you, for example, change the soap that you use if order to wash hands, they will now see this as a potential stress. So obviously, we needed to be-- as we moves as possible as humans from different handlers in order to have as much of an accurate and predictable result. And finally, the disruption needs to be reproducible in the sense that we can do it at will and characterize it in a very specific way. So all this took about four years to develop, but here it is. It's essentially a very simple device. It is a little sweeper that goes from one side of the case to the other. Mice are there, they're happy, and they go to sleep. And during the time that they're about to sleep, all we ask them is to step over the little sweeper and go back to sleep again. But that leads to the arousal, and this is exactly what we're trying to do. So doing their sleep time, which is during daylight, here it is. This is a sleep recording of a mouse. This is DEG. This is DMG. There's no EMG activity now. The amplitude is higher, which corresponds to non-REM sleep. And when you apply the levy, the sweeper, the mouse awakes, goes back to sleep, awakes, goes back to sleep, awakes, goes back to sleep, OK? So now we have characterized the sleep of these mice, and they don't lose sleep. All they do is wake up and go back to sleep, exactly like patients with sleep apnea. Patients with sleep apnea do not have sleep deprivation for the most part. They have a total amount of sleep that is really very normal. The problem is is that they wake up many, many times, so they have sleep fragmentation. And this is work that was done by fellow in the lab, Fahed [INAUDIBLE] who is now back to Israel. But Fahed really started initial experiments. And this is a tumor, by the way. And this is the tumor sleep fragmented. You can see two things. One is much bigger. And second-- and I show this picture because it penetrates into the abdominal cavity, OK? So this is a very aggressive tumor. And this tumor's excised by another postdoctoral scholar in our laboratory, Jamal Jeng, who has now characterized many of those. And you can see that, for those of you, this is sleep control. This is sleep-fragmented tumors. And you can see that the vast majority of these tumors are much bigger, OK? Now, this is, again, the trajectory of the growth of these tumors. It controls sleep and now in fragmented sleep, and it becomes quite apparent that the increase in size or the proliferative process is accelerated in sleep-fragmented mice. So then we asked the second question-- is this just specific to the TC-1 cells? We were concerned that these might be just specific to one tumor, melanoma, or TC-1. So we decided to go and take another tumor, a very well-established tumor model called 3-LLC tumor. And you can see that the results are essentially identical, OK? So sleep-fragmented mice have not only an accelerated progression of tumors. These, by the way, are very aggressive tumors and they grow very fast. But the tumor size is also, as well-- the volume and the size are also both affected. So it became clear to us that the process is potentially applicable to all solid tumors and that fragmented sleep similar to intermittent hypoxia seems to be contributing to the accelerated cancer progression. And I've talked to you about these little fellows called TAMs. Well, we can do very significant counting of these cells. And the way to do it is by doing full cytometry. There are multiple cell types that are inflammatory cell types in a tumor. They're divided into three major groups-- T, T lymphocytes, and they're, of course, subtypes; CD8, which are the good ones that help fight cancer; and then we have T regulatory cells that are suppressors. They're immune suppressors. Tumors that are very big or very aggressive will have very high numbers of T regulatory cells. I will not talk today about T regs. But we are doing a lot of work in relation to this. Second, there are other cells which are mosaically derived stem cells that come from the bone marrow and invade the tumor. And these cells, which are semi-stem, seem to be contributing in a very significant fashion to ulcer tumor progression. Again, this is an area that clearly needs to be studied, and we are proceeding with that. And we're going to go back to our familiar macrophages or little TAMs, because we found again very similar things. We found that the number of TAMs in the tumors that were in mice exposed to sleep-fragmented conditions was also increased, OK? And this is shown here quite eloquently. So there's no doubt that there's the host response, again, that appears to be important. And we wanted to see whether indeed there was any evidence of increased invasion. As I showed you, there were some evidence of that in the flank. But we decided that we were going to do the real test, a much more specific test. And so what we did-- we injected this now into a compartment of the leg. This is a very well-defined compartment. It cannot go anywhere unless it will create a capsule. Then the tumor will be [INAUDIBLE] within that capsule. And this is what happens in sleep controls. You can see this is tumor. And you can see that the tumor is very well-contained inside. And you can see here that the boundaries are exceedingly crisp and defined. And then this is the tumor in mice exposed to sleep fragmentation. So you can see, one, there's really invasion of the muscle of the capsule. And there's almost no distinction of the two. There were some mice that we had to sacrifice, because the tumor went all the way and then ate the bone. So this was a very aggressive behavior of a tumor. And that's something that we really didn't expect to see as much. And so in order to quantify this better, we decided to do some imaging of these mice. Mice that have tumors that are very aggressive will have tongues that release matrix metalloproteinases. These MMPs seem to create the way and open the way for disruption of the collagen, disrupt the protein links, and, by doing so, allow for the tumor to invade. So if we see a lot of activity of MMPs, that would indicate that the tumor is very invasive. And this is exactly what we found. This is an injection of a compound that will become luminescent when MMPs become active. So they will reflect activity. And you can see that in the controlled-sleep mice, there is some degree of activity in the tumor. But the activities markedly increased in mice that were exposed to sleep fragmentation. And this is quantified here in an unbiased fashion through the camera. So, again, what we find is that, one, we have increased macrophages, that these macrophages appear to be much more aggressive, appear to release these MMPs that seem to burst through, and that the tumor then can become as invasive or very invasive and potentially very metastatic. And this is to illustrate that, again, there's a shift to the M2 mold. This is the M1. There's really no changes, but the M2 is markedly enhanced in the TAMs that we see in SF or in C-fragmented MIs, suggestion again that there is a shift of this polarity towards a pro-tumor type of behavior. And we can look at the TAMs, and the TAMs work primarily in the peripheral. These TAMs, they seem to be migrating towards the periphery, where they seem to be releasing all these compounds that are very, very active. And they induce the formation of new vessels. We have stain for a specific marker of new vessel formation. And what we see is the NSF. We can see that there's a lot of new vessels being formed, which indicates a very aggressive phenotype of tumor. This is not the usual behavior of the tumor in normal sleep conditions. So as part of our screening of these macrophages, we found that these macrophages that were different in sleep-fragmented conditions expressed very high levels of an innate immune response marker that has been identified, by the way, by a University of Chicago graduate, who received a Nobel for his discovery of innate immunity. And these are toll-like receptors. And among the toll-like receptors, there are many, many. It's a very large family. It's a super family. But this one, TLR-4, that is particularly involved-- and has been associated also with cancer-- but with other features. And in this case, what we found was that TLR-4 was highly expressed in the macrophages of sleep-fragmented mice. And when we looked at TLR expression of these macrophages, you can see that in the protein, it's also highly expressed. But then we took advantage of a TLR knockout and obviously confirmed that they don't have it. And interestingly enough, when we exposed TLR-4 for knockout mice to sleep fragmentation, there was no difference in the growth between sleep fragmentation and control, suggesting that indeed this induction of TLR-4 seems to be a very important facet of the increased aggression and invasion of the tumor and that maybe through modifications of TLR-4 mediated pathways, we can maybe change the tumor behavior and this could be yet another target for cancer therapy and, in fact, one that is actively explored by some groups now in Europe particularly. So sleep perturbation, similar to those in sleep apnea, can accelerate tumor growth and invasiveness through TAM recruit and TLR-4 signaling pathways. And hopefully there will be much more to come. So I'm going to shift very quickly now to some clinical so that I hope will convince you that there is a biological possibility to something that really presents a clinical problem. So there aren't many studies, but you've already about the Wisconsin cohort. And so Javier Nieto, who happens to be from Spain but is now the chair of epidemiology at the University of Wisconsin at Madison, decided to look into cohort and to look at cancer mortality. So they took the cohort. They all have sleep studies. They knew who had cancer, and they look at cancer mortality. And these are their results. Is a court of 1,522 patients, subjects from the community. And as the severity of sleep apnea increased, the best from cancer, as I alluded to you in a heated fashion now quantified in the right way with appropriate epidemiological tools seemed to be increased by the presence of sleep apnea, particularly when sleep apnea was more severe. And this is part of the mortality adjusted. What they looked was-- first of all, there was clearly an increase with severity. You can see this. Then the severity of hypoxia was also a major contributor. And they found that if you were in the percentile down or the centile where hypoxia was the most severe, that the risk was actually multiplied by eight-fold or so, so in an adjustment model. So suggesting that indeed intermittent hypoxia, they did not look at the sleep quality. They just looked at intermittent hypoxia, was really associated with the same features that we see in the animal model. So there is an association between sleep apnea and risk of cancer mortality in a population-based sample. And we need to understand much more, whether this is related to obesity, interfaces with obesity, and so on. So much more work to do. So a second group-- and it's a group that I was fortunate enough to participate in the formation of this group many years ago, and they've done an outstanding job at getting together as a country, establishing multiple sleep laboratories across the country, and then do joint trials. And this is the Spanish trial group. They have been very involved in sleep apnea and cardiovascular. And they decided to look at the cohort, which is very large, and see whether they have any signal in relation to cancer. So this is a group that started collecting here between 2000 and 2007. They finish in 2010. These are the hospitals the participated. You can see they're all over Spain. And they have 5,400 patients. And what they found was that of these they had 5,400 that could be analyzed. And of these 5,400, they had 369 deaths. And of those deaths, 24% of those were from cancer. Again, a very shift towards a high risk from cancer in a population with sleep apnea. And when they looked at all the variables in their analysis that all the things that need to be adjusted for, that have been defined as either risk for cancer or associated with sleep apnea. It is on the list. And they did tertile separation, and they found that indeed the mortality when you have more severe sleep apnea, it was higher. And the cancer mortality, as shown here, was located-- the cancer was located primarily in respiratory GI and genitourinary tract solid tumors, OK? So this is quite interesting. We don't see lymphomas and leukemias, so we don't see metalogical malignancies. We see solid tumors emerging as the major driver for deaths in a population of patients with sleep apnea. And a severity dependent increased in the mortality as the severity of sleep apnea increases. And this is shown here both as a function of the AHI. You've heard about this measure. It's not as significant, but certainly associated with a degree of desaturation. So the interesting thing is that in this cohort, which is much bigger, they found that women and very old patients were actually somewhat protected. Now, this is very interesting, because elderly could be protected by the survival effect. They have already bypassed the risk. So if you're already so old, you're going to be able to survive, because you've got the survival genes. You are super-old. But women is a very interesting one, because it can also imply that it's estrogen or at least some hormonal components may interface with sleep apnea, and we know that they do. And they could modify risk, particularly in younger women and so forth. So the conclusion from this study was that sleep apnea severity, similar to the one from Wisconsin as measured by the coefficient of the saturation at 90% or below 90%, was associated with cancer mortality. And this association in their cohort in Spain seems to be more linked to men. None of these patients are those under 65 years of age. But what about incidence? And in no way do the animal studies that we have done addressed the issue of incidence of cancer. To do that, we need to use very different animal models, and we are planning to do those hopefully in the future. But nevertheless, this is being investigated by, again, the Spanish cohort. They looked at whether there was an association between obstructed sleep apnea and cancer incidence in this large cohort. So these are, again, some of the centers that participated. You can see that come participate in some, but not others. They collected a total of 4,900 subjects or patients with sleep apnea. They were analyzed over a period of 4 and 1/2 years, and they looked at the incidence of cancer in this population. So again, the variables that would control the separation into tertiles. And these are the tumors that were found. Again, solid tumors-- breast, lung, prostate, and colorectal. And the same thing appears to be the case. So the more severe the sleep apnea, the higher the incidence of solid tumors in this population. One, increased incidence-- this needs to be reproduced in subsequent cohorts. I know that studies are ongoing and including the US for cohorts that have been available. But second, progression of mortality seem to be also associated with sleep apnea. And by the way, again, the issue of severity of intermittent hypoxia seems to be driving, at least in theses cohorts, both the incidence as well as the mortality. I'm going to skip this for the sake of time. All the multivariate analysis confirm all these things. And again, what they find in these Spanish courts is that this appears and these findings appear to be limited to men and to patients younger than 65 years of age. So there's a discrepancy of genders and also a H protecting effect when you reach an age of about 65. And by the way, this is true, too, for cardiovascular morbidity or mortality. The patients above the age of 65 with sleep apnea do not have an increased death rate from cardiovascular disease. But between the ages of 45 to 60, they have a very significant and very major increases in mortality during that group. So maybe there are differences that have to do with survival or with hormonal changes. So what are my final comments? And I will end here. One, I think that I've shown you some data that there is a biological plausibility that can link sleep apnea to cancer. And I think that that is quite firmly established by the results that I've presented you from our animal models. And what's most important now is that we need really to zero in on many of the mechanisms in order to use that as a model that will serve hopefully in, one, prevention and treatment and understand much better what areas of sleep we need to address in patients with cancer and how to prevent, whether to improve their prognosis or to prevent the occurrence of cancers altogether. And second, that we have already preliminary data from patient cohorts that indicate to these risks and that undoubtedly there's a need for prospective studies in order to both address histological classification. As you know, the heterogeneity of tumors is now a very important facet of prognosis. And the organ location-- why are these tumors particularly located in the areas that I showed you, as opposed to others? And what happens if you treat sleep apnea? Will you reduce that risk? That, too, needs to be established so we would reduce immortality. So I'm going to thank now to the Spanish group. This is [INAUDIBLE] here. But more importantly to many of the most important collaborators and friends that we have here at the University of Chicago-- Isaac, Fahed, [INAUDIBLE], Lev Becker, and Jamal. And I didn't have a picture for Shelly and for the others, but thank you for your attention. And I wanted also to mention that two of our collaborators, [INAUDIBLE], were really instrumental in getting us to learn how to do the lung tumor model that I've presented to you since we were very, very naive at the beginning in anything that related to oncology. So our thanks to them as well. Thank you for your attention. I'll be happy to answer any questions. [APPLAUSE] SPEAKER 2: Thank you very much for this wonderful lecture. Questions. We have a good five minutes. SPEAKER 1: Yes. SPEAKER 3: It's something of a chicken-and-egg question. You're suggesting that sleep disruption and so on is a risk factor for inflammation. Does it go the other way around? If you have an inflammatory condition, does that predispose you to sleep disregulation? SPEAKER 1: Absolutely. This is a great question, the chick and the egg. And the chicken and the egg coexist. So both directions are applicable. Jim Kruger-- when Jim Kruger was a graduate student in Pappenheimer's lab at Harvard, Jim came through a very interesting observation-- is that when the infected mice with influenza-- he found that these mice tended to sleep much more and ended up discovering IL-1. And through that process, he became certain that many cytokines-- since then, there's been a lot of work in the last 30 years. And the whole career of Kruger and colleagues were multiple factors, such as cytokines and inflammatory mediators do modify sleep and disrupts it. They can alter sleep structure, sleep architecture. They can disrupt the sleep continuity. Or they can enhance certain types of sleep, depending on the nature of these. And these are quite interesting, because these would get to the issue, what happens when you have disrupted sleep related to inflammation. Is this really the inflammation that may exist, for example, in Crohn's disease and so on? Could it be that it's also affecting sleep? And both of them seem to be much more effective in inducing oncogenesis or tumor proliferation in those cases. SPEAKER 3: Do tumors [INAUDIBLE]? SPEAKER 1: There's no evidence of two tumor, per se. There's tumors that release certain cytokines. As you know, cachexia has been the initial findings of TNF-alpha. This is how it was described. It was called cachectin at the beginning. Since then, obviously we know this tumor necrosis factor alpha. TNF-alpha is a major potent, a very potent somatogenic factor. Cytokine, when injected into areas that they have to control wakefulness, it will induce sleep and will increase sleepiness vice versa. In areas that generate sleep rhythms, it will enhance non-REM sleep. So we know that fragmented mice, for example, also have increased release of TNF. And if you block that TNF, you can actually take away the [INAUDIBLE] of these months and yet it won't affect the tumor. So I think that there's a variety of perturbations that may have a general affect-- behaviors such as sleep and so on that relate to cytokines or mediators of inflammation. But that some of them, the local effect may be much more prominent than the general effect. And the section of the two will be very important in order to identify potentially ways of targeting locally but without affecting the behavior of the [INAUDIBLE]. SPEAKER 4: Great talk, David. Two questions. And again, it may sound bizarre, but I'm asking an adult pulmonologist. But patients with COPD who may have hypoxia-- now granted, they may not have intermittent hypoxia, but it maybe just be hypoxia throughout the day. Certainly COPD is a risk factor for lung cancer. But I'm curious if there was any data since these studies have been coming out with adult COPD-ers who need oxygen or hypoxemic have higher risk for tumor development or tumor growth, compared to those COPD-ers who don't have hypoxemia. SPEAKER 1: OK, so that's a tough question to answer, because there's really no specific answer to this. The question is whether COPD is a risk factor. It is, particularly for lung cancer but also for colon cancer, prostate cancer, and pacreatic cancer, and in women for breast cancer. But there are some confounders. The first confounders is that many of the patients with COPD have smokes. So now you will have to exclude all of those in order to have a more pure population of patients that have not smoked and yet develop COPD, which exists of course. But then you're much more restrictive in your analysis. So I'm not clear. Second, it would have to be that these patients are treated and would be adherence versus not adherence to the supplemental oxygen. So it makes it complex to start looking at the average area under the curve of situations and then try to link it. There's a little bit of that is ongoing now in [INAUDIBLE], a cohort in France. The group in Paris, which leads the [INAUDIBLE] coordinated group for all France and has data now on supplemental oxygen in COPD for the last 45 years, where they've done home management and continuous monitoring. And now they do it through the telephone. So they have oxygen trends once every three months in patients throughout-- they start looking at this. So I hope that they will be able to look at their longitudinal data and start extrapolating some signal. But the preliminary findings-- at least what they report to me when I was there with them-- was that they seem to find that COPD overlap syndrome is associated with increased tumor incidence and adverse prognosis. And the overlap syndrome, for those of you who don't know, are COPD will OSA. So sleep apnea with COPD, which can coexist. And second, the patients that refuse to oxygen or were not adherent seem to have also an increased incidence and adverse prognosis. But these are very preliminary analysis on about 240,000 patients. So it's a very large goal. So we'll see a little bit more as they finalize their study. SPEAKER 5: I was wondering if [INAUDIBLE] hormones like [INAUDIBLE], which was induced for-- in terms of hypothesis [INAUDIBLE] sleep apnea. Quick help for controlling for the level of stress. SPEAKER 1: So we have measured corticosterone, which is a reporter. The question is whether cortisol as a matter of stress could help in understanding a little bit more of the mechanisms. So two things. One is that the first observation is that we have measured in the animal models corticosterone, which is a similar way of measuring cortisol. And we found no alterations in sleep fragmentation. Now, that is not true in intermittent hypoxia, because in intermittent hypoxia, we use quite a significant severe model. So severe model in intermittent hypoxia will lead to alterations of cortisol. Now, that said, patients with sleep apnea do not have very significant alterations in corisol. And what they have is a shift in their peak, which is interesting, because it seems to be one of the few manifestations of circadian disturbance that is manifest as a shift in the circadian beak of one melatonin and two cortisol. And now, we know the melatonin can be a protector against cancer. And it is likely that this shift may also have implications. But I don't think that anybody at this point has a clue as to how this shift, not elevation, but actually shift in the beak affects the any way the tumor associated behavior in cancer. Yes? SPEAKER 6: That was a very intriguing talk. Sir, my question is is about something that I noticed-- the difference between mice model and the humans. So you see the effect in mice very impressive in hours, up to 12 hours. And in humans, what you show it takes months even probably more, more than months and years. Respectively, do you-- could speculate with additional mechanism in humans that affect the tumor and growth and [INAUDIBLE]? For example, the epigenetic factors, which you can define some windows, then, to reverse that process in humans, because obviously it takes much, much longer time to see the effect. SPEAKER 1: Sure, [INAUDIBLE]. Of course, it has to be epigenetic, because otherwise you wouldn't be asking the question. So the question is really not ours. First of all, the animal model is not intended to recapitulate all the faces of oncogenesis. The first thing that we don't do is we introduce a cancer with a certain load already into the mouse. So it's not the same process as the continuous formations of single cells that undergo malignant transformation and are cleared by the immune system. So to some extent, it is possible based on the literature of the human data that patients with sleep apnea may not be able to remove as efficiently tumor cells that occur in a natural environment, are identified as foreign cells, and then the immune-innate system is harnessed to destroy them. And it could well be that the patient will sleep apnea because of the shift in the polarity of their macrophages or many other immune-related factors may not be as adept at doing so. But I wouldn't still-- doesn't mean that the-- the second is not in ours, because the mouse model takes about a month or a month and a half. So it's still an extended period of time in order to see it grow. But the in vitro, of course, is a very fast process of two to three days. That's about it. Or fours days. The second part-- one does not negate the other. So is it possible that intermittent hypoxia or sleep fragmentation modify the epigenal? And the answer is yes. There's evidence that we have generated in our lab, as you know, that the children, even children with sleep apnea, have already epigenetic marks that are altered in the context of the sleep apnea. And we have Layla-- Dr. [INAUDIBLE] actually shown very nice that it correlates with vascular funtion and other fenal types. So it is very likely that these epigenetic phenomena may be persist or play a role. But it's up to us to want to identify those. And I hope that both you and Dr. Cortez will invest quite a substantial amount of energy in doing so. But second, whether they play a role in which side of the process? One of the actual oncogenesis, the formation and perseverance in the immune function. In the proliferation, in the invasion, it will be very important to identify which phase of the constellation of the tumor life they play a role and whether we can intervene to that effect. All this is very speculative, but it's nothing better than to speculate so that we can leave free right to the imagination. That's great. All right. Thank-- oh, sorry, go ahead. SPEAKER 7: I want [INAUDIBLE]. SPEAKER 1: Well, there's a very, very interesting question and one that is being investigated. So the question is whether circadian rhythm disruption can affect cancer. There's a very large body of literature linking circadian clocks and cancer biology. And we could spend the next several hours talking about how specific local tissue clocks have to do with specific disruption of cancer biology and immune function as well as central clocks than having an effect on top of it and superseded. And those have been studied quite interesting. And this actually was triggered by a very intriguing observation that depending on when you did the circadian phase, you injected the chemotherapy, there seem to be differences in the outcome. This was done in a very large cohort of the NCI, and they identified that patients were injected in phases that were specifically circadian-related that there was improved outcomes, compared to if they were done at random in the circadian phase. So that led to a major effort over the last 15 years or so and circadian clock biology and cancer. And there's a plethora of literature, but I won't be able to discuss it in detail today. OK. Well, thank you all for coming. And have a nice weekend.
