Dr. Robert Vessella Video (Text Version)
Title of Talk: Dissemination of Prostate Cancer Cells to Bone: Implications for Disease Progression and Tumor Cell Dormancy
Speaker
Thank you Dr. Thompson for your excellent and informative comments on GLIPR-1. The Prostate Cancer Research Program has been placing emphasis on proposals to develop and utilize new biomarkers as a focus of the award mechanisms. In recent years, there has been more and more research on circulating tumor cells and disseminated tumor cells as a major of prognosis or status in the advanced cancer patient.
The–the distinguished Dr. Robert Vessella will be presenting next on CTCs, DTCs, tumor cell dormancy, and activation. Dr. Vessella is current Professor and Vice Chairman of the Department of Urology at the University of Washington Medical Center. His research interests include genitourinary cancer, biomarkers, bio-specimens, metastases and xenograft models of prostate cancer. He received–received his Ph.D. in Microbiology and Immunology from the University of Mississippi Medical School in 1974; the title of Dr. Vessella’s talk is Dissemination of Prostate Cancer Cells to Bone, Implications for Disease Progression, and Tumor Cell Dormancy; Dr. Vessella.
Robert Vessella, Ph.D.; Professor and Vice Chairman, Department of Urology, University of Washington Medical Center
Thank you for that nice introduction. And I want to start by thanking the organizers, Natasha and all the others for inviting me to this prestigious meeting. It’s been a lot of fun so far and we have another day and a half, so I’m looking forward to it. I also want to thank the DoD for funding this research. Very early on this was considered to be extremely high-risk and the DoD stepped forward and funded us, got us going in the right direction, and it’s been a lot of fun ever since, so thank you to the DoD members and I really, really appreciate it.
Okay; let’s see what we got here. Okay; the enigma of prostate cancer as we see it from our eyes in my laboratory is that there’s a very high propensity for prostate cancer to metastasize to bone. There was specifically a bone session yesterday and there was also a metastases section yesterday that emphasized this point–very high proclivity for metastases to the bone. And in contrast to many other human tumor types the–when you do an autopsy on these patients, at least a half of the patients show minimal clinical relevant disease in their visceral cavity; almost all the disease is in the bone. In fact, about 90% of the patients with advanced disease have skeletal metastases. So without the big visceral burden that you see with other types of disease, it begs the question, why do these patients die?
Also, prostate cancer in contrast to other cancers that do metastasize to bone perturbed the normal bone remodeling and they perturb it in a manner that results in what we call the osteoblastic response. This is the response of bone when you get lots of new bone growth in comparison to a metastases such as from breast cancer where you get an osteolytic event which is degradation of bone. Now I’ll say from the start that you end up in prostate cancer by having both events occurring simultaneously and you do have oncolytic events as well as the osteoblastic effects, but the predominant view is that it’s osteoblastic.
Now this is an example of that osteoblastic effect. This is a [inaudible] body from one of our rapid autopsy patients and you can see I hope here–there you go; you can see that there’s disorganized, lots of new bone and the bone marrow had been totally replaced by tumor cells. So this is what we’re dealing with and this causes a very high degree of morbidity and eventually mortality in these patients, but it can be extremely painful when you have metastases of this nature to the bone.
Now in our lab we arbitrarily have broken down metastases to the bone which is kind of our little niche into three phases of study. The first is trafficking and dissemination of the tumor cells; how do they get to the bone? Is there a specific targeting; is there an attraction for tumor cells to the bone or is it simply a–a physical means of entrapment that they get into the capillaries, they get trapped, or is it again, an–something to do with specific adhesion and specific targeting? When do they disseminate and get to the bone? Is it early in the disease or only late in the disease process? And what are the characteristics of these disseminated tumor cells when they get to the bone? Which cells actually get out of the primary tumor, circulate, survive that process, and end up in the bone?
Then once they get to the bone they have to adapt, they have to grow, and then they perturb this remodeling aspect in the bone. What adaptive events occur? And we have a process called the vicious cycle where tumor cells in a new environment produce factors that influence the cells in that environment and then those cells in that environment produce factors that influence the growth of the tumor cells. So it’s a vicious cycle. And what are the factors produced by prostate cancer cells that are unlike the factors produced by breast cancer cells or myeloma cells that cause this new bone growth and again the classical osteoblastic response?
Then another area that we’re highly interested in is how do we model this? How do we develop models in mice that would mimic the same process that’s occurring in man? Finally, the clinical manifestations–how are the characteristics–what are the characteristics of clinical bone metastases? What factors again are produced by prostate cancer cells that cause cachexia? Cachexia is a wasting syndrome, muscle and fat is degraded and patients to go through a wasting and you will have a patient that is normally a 200–240-pound patient that when he comes to autopsy he’s 70, 80, 90-pounds. He has basically wasted away. What are the factors, the cytokines that are involved in that? And I won't get into that today but a very interesting phenomenon.
And then last, tumor cell dormancy, we’re going to talk about this at the end of my talk.
So initially we’re going to talk about trafficking and dissemination. We can't cover all the items there; we’d be here all morning but we’re going to talk about dissemination and shed tumor cells and then a little bit about dormancy. I’m going to give you an overview; I’m not going to give you a lot of hard data because we only have a few minutes to talk but hopefully it’ll be enough to stimulate a discussion within the next hour.
So for this talk the definitions are as follows. You have heard of circulating tumor cells and you have heard people use disseminated tumor cells. For my talk, the way the scientific community is now going is that circulating tumor cells are those shed tumor cells that are found in the peripheral blood. Disseminated tumor cells are those tumor cells that end up in the final place of rest, so a–a shed tumor cell that ends up in the bone marrow for example would be called a disseminated tumor cell. If it’s isolated from the blood and still circulating it’s circulating tumor cell so that’s how my definitions are going to go during the day.
And some of you saw this same illustration used yesterday in the metastases session. Basically you have tumor cells coming out of the primary tumor. They disperse through either the lymphatics or the peripheral blood. In this process, if I were to isolate tumor cells here, they would be considered circulating tumor cells–same up here, and then once they get to their distant organ they’re termed disseminated tumor cells or DTCs.
Now from our rapid autopsy program we’ve been able to determine the prevalence of tumor cells grossly and microscopically at various bone sites. And you can see from this illustration that it really is a systemic disease. The tumor cells disseminate to all areas of the skeleton; there’s a very high frequency and in general about 70% of the bone sites that we take at rapid autopsy and we take 20 different pre-defined bone sites, vertebral column, the ribs, all along this skeletal region, about 70% on general we’ll find tumor cells, so it is a systemic disease within the skeleton.
So the premises that we’re dealing with is that all metastases emanate from disseminated tumor cells or shed tumor cells. And initially these are from the primary tumor. Later on in the disease process, one can argue probably quite effectively that metastases shed tumor cells and that these tumor cells seed other metastatic sites. But initially they’re all coming from the primary tumor.
We’re hopeful that the study of these cells can provide valuable insight into the metastatic process overall, provide information about tumor cell dormancy, and also give us some information on prognosis and in monitoring response to therapy. So what are the fundamental issues? Well again, all metastases emanate from DTC but not all DTCs cause overt metastases. So even though we can find disseminated tumor cells in a patient’s bone marrow, it does not necessarily mean that–that patient is going to develop a metastases or clinically relevant metastases at that particular site. And we’ll get into why in a little bit.
So there is dormancy; these DTCs can become dormant. We don’t know if that’s an initial event or if it’s an acquired event. But these DTCs can seed the bone marrow and sit there for 5, 10, more years and the patient appears after radical prostatectomy to be perfectly cured, all of the symptoms–I mean there are no symptoms, all the clinical signs are that the patient is cured and then he’ll come back into the clinic, many, many years later with a metastases and this will be after radical prostatectomy so the primary tumor is long gone. It’s been pickled; it’s been given to the lab. But yet, this patient is still harboring cells in the bone marrow for a long period of time.
Can we define and apply molecular characteristics of CTC and DTC and differentiate between those cells that will or will not adapt and thrive at these bone sites? Can we also apply this insight to determine what breaks dormancy? Dormancy is absolutely fascinating when you think that these cells are sitting there for maybe 5, 10, 15 years, but what breaks dormancy? What then stimulates these cells to begin dividing and cause a clinically overt metastases that the medical oncologists then have to deal with? And can we learn from our studies of CTC and DTC about detecting these cells and using the information gleaned from the detection to develop assays that could help perhaps in determining whether a therapeutic reaction is favorable or not favorable to the long-term therapy of that particular patient.
So there are many methods for detecting and isolating these CTC or DTC. We’ve used this particular method for a number of years. There are many more coming aboard all the time; there are advantages and disadvantages to each, but this is the one that we’ve used for years. It’s a home brew assay and I’ll go through it with you at this time. So if we start with cells from the bone marrow, starting with a bone marrow aspirate typically you know seven and a half to ten ml of a bone marrow aspirate, we first do a negative selection to get rid of anti-CD45 and anti-CD61 cells. These are using para-magnetic beads. We target initially CD45 and CD61 and then we go through a positive selection at–looking at HEA125 which is an anti-EPCAM. EPCAM is an epithelial marker on the surface of the cells.
If we start from cells in the blood, a peripheral blood sample then we can bypass the negative selection process and go directly to the anti-EPCAM, remove these cells and study them, and what’s important at this point is that we have anywhere from 1,000 to 10,000-fold enrichment of the tumor cells but it’s not a homogeneous preparation. There is no method out there which will give you a homogeneous preparation of tumor cells at this particular point in time. What is important however is when you finish this process you end up with viable tumor cells, and the way you do that is that you have a wet mount and using a second antibody to a different epitope of EPCAM. You can highlight the cell and again it’s in a background of other lymphocytes so we get an enrichment but not homogeneous preparation. And then you can go in with a micro-pipette and literally suck up one cell at a time and put them in a vial. And God bless my technicians who are cross-eyed and wandering around aimlessly who have done hundreds of these in the dark room and it’s really remarkable of their persistence in the ability to do this.
But what you end up with is a homogeneous–well I don’t want to say homogeneous; you end up with a population of cells that you believe are the tumor cells in a vial. So now you can study these tumor cells using RNA extraction. You put them in culture, put them in mice; whatever you want to do–they are viable cells.
Now depending upon the method you use for a detection the frequency of detecting CTC and DTC upon initial presentation ranges from about 20% to 88% in the literature. In our particular situation and looking at again bone marrow aspirates, we’ve looked at greater than 900 and throughout the course of the disease the range is anywhere from 40% to 70%. So the frequency of detection is quite high even in those patients who initially present to the clinic; even if that patient has localized disease we can still detect the disseminated tumor cells in their bone marrow. And we’re going to go through some of this in just a second.
So about 50% of patients, these are our data, about 50% of the patients pre-radical prostatectomy have detectable DTC and this is in the bone marrow. You get a high percentage in patients who have metastatic disease as you would expect. And when you have a disease-free interval after radical prostatectomy when the patient has no evidence of disease who was considered NED, we can still find 45% of the patients after 5 years who still show the presence of these DTC in the bone marrow. And we consider these to be dormant. These patients where we can detect these dormant cells have a sevenfold higher risk of recurrence at some point in time, but again not always–it’s just a higher risk.
Now the mere presence of DTC prior to a radical prostatectomy did not again necessarily mean that patient is going to have an overt metastases or a poor prognosis. Now we need to get away from just looking at frequency of detection–a yes or no–to looking at the character of these particular cells. But characterization is really a challenge because in this process of isolating the cells we can only isolate one to ten CTC from the blood and maybe 10 to 20 from the bone marrow. That’s a true challenge and it challenges the current techniques for genomic studies such as expression arrays or comparative genomic hybridization.
Now in advanced disease you can get into the thousands; that’s possible. But generally again you’re dealing with a few dozen cells at the very best, so characterizing these cells becomes a real challenge. Our particular project was to isolate these DTC, compare them by back array CGH, comparative genomic hybridization and do CDNA microarrays with Pete Nelson over at the Fred Hutch. Then to compare the profiles that we see from these genomic studies of the DTC and compare them to the radical prostatectomy sample and also compare them to metastatic foci.
Many of you know and have seen throughout this meeting that there are numerous genomic aberrations in prostate cancer. Here are some that are relatively well known. Using array CGH you can determine where the genomic alterations occur. This is not that difficult when you’re dealing with a metastases; in this particular instance–this is breast cancer and you can see a gain–these are chromosomes across the bottom–a gain and some losses. But when you’re trying to deal with 10 cells, it becomes a technical challenge. We worked with Barb [Trask] over at the Fred Hutch Cancer Research Center and her graduate student Ilona Holcomb to overcome this and have really been very pleased with this data. We’re now moving on and doing SNP arrays but let me show you from the array CGH data.
In this particular slide, you can see these are patients from–54 patients where we’ve done comparative genomic hybridization and you can see that we have some patients only in their–in their DTC they have one aberration, some have two, and some have three. Now compare that to what we find–this is at the time of radical prostatectomy so this is early in the disease–compare that to this profile and these are the same DTCs, same method used from patients with advanced disease. And look how chaotic that pattern is. So what it tells us is that there is an evolving genomic aberration that occurs in these patients over time; they start with a few and over this long period of time that the cells are residing in the bone marrow they acquire other aberrations until they become totally chaotic.
Now looking at metastatic prostate cancer, you see the same pattern–the gains and tremendous numbers of losses. Look how chaotic that chromosome is in this particular metastases. This paper was published back in 2008 and as I said, Ilona Holcomb is the first author. So the result from our studies is that there have been actually very few comparative studies because of the challenges to characterize just one to 20 cells. In localized prostate cancer where we had 54 patients that we looked at comprehensively, 68% had genomic aberrations and typically these were one to three. And it suggests that they’re a partial epithelial mesenchymal transition from our gene array data which I’ll get into in just a moment.
In advanced prostate cancer, we looked at 11 of these; all the patients had these chaotic genomic events. And again it–it appears to promote the concept of an evolving genomic evolution as these cells sit in the bone marrow.
Now I mentioned EMT, epithelial mesenchymal transition. This is a–a somewhat controversial area in whether EMT actually occurs in vivo or whether it’s more or less an in vitro artifact. But again the cells disseminate, and the thought is for these cells to egress–get into the blood and egress and form a tumor they need to become more modal and more plastic. That is a process of EMT; they lose their epithelial markers and they gain mesenchymal markers. Then once they establish that niche in their distant site, stromal cell interaction begins to take over and then it goes back to MET, mesenchymal epithelial transition, and they regain their epithelial markers.
In our data we’ve been able to–in our cells we’ve been able to take these disseminated tumor cells from either radical prostatectomy patients at the time that they have no evidence of disease or in advanced disease. And these are some of the epithelial markers and you can clearly see that as we go this way they have lost a number of their epithelial markers, and in advanced disease they have gained back their epithelial markers. Other than S100A4 we don’t see the opposite pattern as we would expect if this were truly 100% EMT. But you do see at least in S100A4 where you have up-regulation early and then a loss late.
Now you have to remember that these are pools of cells. These are 10 cells that are being analyzed, so it’s an average and that is one of the drawbacks to our studies and other studies; we need to get down to the single cell level to be really accurate in what’s going on with these disseminated tumor cells. But it at least suggests that there is a partial EMT going on with the loss of the epithelial cells.
Now tumor cell dormancy; now both in breast and prostate cancer, metastases usually occurs in the bone as I’ve mentioned. And in some of these patients metastases aren't evident for 5, 10, or even more years. Yet in our series we were able to show that approximately 50% of patients who were in this NED period actually had detectable DTCs in their bone marrow. And we have at least two or three patients now in our clinic who are between 15 and 20 years post-radical prostatectomy, and we can consistently detect these DTCs in their bone marrow. They are dormant; the patient feels fine. PSA levels are great, but these cells are still detectable.
And again this long latency period is considered tumor cell dormancy. Here are some of the data and you can again see it’s around 50% greater than 5 years.
So what differentiates an aggressive DTC from one that is dormant? Is it the cell or is it environment? Now we know that there’s heterogeneity in this population, and it probably is going to require us to get down to the single cell level to be able to distinguish the aggressive cell from the indolent cell because we are dealing with a mixed population. We’re working on being able to get down to the single cell level. I think we’re very close to being there but it’s a huge challenge. Do all patients have a sub-population of dormant cells? And are these dormant tumor cells similar to stem cells? One of the problems that we have in prostate cancer is that we don’t have biomarkers that define what a dormant tumor cell should be. We don’t have defined markers of what a stem cell should be for prostate cancer, so being able to determine this is going to be a challenge.
And again, very important–what induces dormant cells to become aggressive and begin to grow out. Some of the mechanisms of dormancy are oncogene inactivation, lack of androgenic switch, lack of critical growth factors that at the site of seeding, active metastases suppressor genes. Cary Winker Shaffer and I hope to be establishing a collaboration in the near future to look specifically at this; this is a very hot area and there’s strong possibility that some of these cells have metastases expressive genes that are active that keeps them from growing once they establish at the distant site. I told you about the evolving genomic patterns. Is this an aspect that we need to consider? Aberrations and adhesion factors, loss of epithelial stromal cell cross-talk and of course the all important immunological factors.
Immunological factors could inhibit tumor growth at the metastatic site but at the same time inflammation could promote tumor growth at a metastatic site. So this is a double-edged sword.
So from our conclusion of our DTC studies, we do believe that the study of CTC and DTC should provide considerable insight into the whole process of metastases and dormancies. And we believe that opportunities will develop from the study of these to develop new assays and to shed insight into the whole metastatic process and possibly monitor therapeutic response as is incurring in some therapy trials right now looking at rises and declines of circulating tumor cells in the presence of therapy.
Now it’s very important to remember again that heterogeneity exists in this disseminated tumor cell population and in the CTC population. And I’m going to conclude with three quick slides highlighting this particular aspect.
Several people have asked with our rapid autopsy program, do we see differences between soft tissue mets and bone metastases and clearly this is work that Pete Nelson helped us with in looking at heat maps. And I’m not going to go over this but these are bone mets; these are soft tissue mets and you can clearly see in this selection of approximately 70 genes the differences between bone mets and soft tissue mets. So the difference is they obviously do exist.
Now what’s even more striking and this is an older slide but I just want to bring this message home; when a patient has bone metastases and you look at enough bone metastases in a given patient–not one or two but as I mentioned, we acquired 20 different bone sites from each patient–and you begin to look at marker expression, here’s PSA. These are the patients. If you look at the percent of PSA-positive cells in the metastases, you certainly see quite a few number are 100% just like you would see in the primary tumor. What you also see in the same patient 0%, 10%, whatever, so when–within one patient the expression of a given biomarker in their bone mets can be extremely heterogeneous. We’ve heard previously this morning about the androgen receptor. The same pattern–okay; 14 different patients; some of the sites have no expression; some of the sites have very high expression. But this again is all within a given patient. So you need to be aware of that when you’re beginning to look at targeted therapies in metastases.
Metastases are extremely heterogeneous as Angelo has said; there are genetic aberrations, gene expression differences, protein expression differences, and obviously therapy response differences. So as you move ahead, if you’re going to design targeted therapies with the hope of treating metastatic prostate cancer, you can't base all of your data on the expression of a given biomarker in the primary tumor because chances are it’s going to be differently expressed in the different metastatic sites in the patient. And you need to be aware of that or you’re going to be going down a path that’s going to lead to basically nowhere. So with that I’ll conclude and I guess we have a discussion session.