FDA Home Page | CBER A-Z Index | CBER Search | Contact CBER | CBER Home Page

DEPARTMENT OF HEALTH AND HUMAN SERVICES FOOD AND DRUG ADMINISTRATION UNRELATED ALLOGENEIC CORD BLOOD BANKING AND TRANSPLANT FORUM Cosponsored by: Center for Biologics Evaluation and Research, FDA and National Heart, Lung and Blood Institute, NIH Tuesday, August 15, 2000 8:00 a.m. Masur Auditorium Building 10 National Institutes of Health Bethesda, Maryland C O N T E N T S Session 3: Transplant Outcome Analysis Comparison of Related BMT and UCBT, Dr. Mary Horowitz New York Blood Center, Dr. Pablo Rubinstein St. Louis Experience, Dr. Donna Wall Consolidated Data of Duke and Minnesota, Dr. John Wagnerd Adult Experience, Dr. Mary Laughlin Japanese Experience, Dr. Tsuneo Takahashi Discussion:       Can We Define an Acceptable Unit for Transplant? Session 4: Ex Vivo Expansion Expansion in Bags, Dr. Ian McNeice Expansion in a Bioreactor, Dr. Joanne Kurtzberg Stromal Factors, Dr. Catherine Verfaille Discussion:       What Do We Need to Know? Session 5: What is a Stem Cell? Introduction and Overview, Dr. Hal Broxmeyer Comparing Umbilical Cord Blood and Bone Marrow Stem Cells, Dr. Esmail Zanjani Potential New Markers: ADHase, Dr. Clay Smith Potential New Markers: FRIL, Dr. Jeff Moore Importance of the CFU-Meg Assay, Dr. John Adamson Correlation of CD34 Positive Cells with CFU, Dr. Jan Visser Discussion:       Correlation with Outcome Brief Summary and Future Directions, Dr. Joanne Kurtzberg PROCEEDINGS Session 3: Transplant Outcome Analysis Comparison of Related BMT and UCBT DR. HOROWITZ: Good morning. I am Mary Horowitz. I am the Scientific Director of the International Bone Marrow Transplant Registry, and I will be starting off this session on transplant outcome analysis. It has been historically true in the field of hematopoietic stem cell transplantation that most allogeneic transplants involve HLA identical sibling donors and, in fact, much of what we know about transplantation clinically derives from that experience, and experience has shown us that most of the principles involved in unrelated donor transplantation are very similar to the principles involved in related donor transplantation. I think it is probably appropriate to start off this session which will, of course, focus on the results of unrelated donor cord blood transplants, with an analysis of the results of related donor cord blood transplants. Could I have the first slide? Now, that is easier said than done because there really aren't a lot of HLA identical sibling cord blood transplants done, and the reasons are obvious. To have an HLA identical sibling cord blood you need to have a pregnant mother at the time that you develop your indication for transplantation. For obvious reasons, that is most likely to happen if you are a child with a disease that requires a transplant and, even more so than in the unrelated donor setting, the overwhelming majority of HLA identical sibling cord blood transplants are done in children. Even so, numbers are small, and to be able to look at this issue it was necessary to combine data from both the International Bone Marrow Transplant Registry and EUROCORD. This was done in a study that was recently published in The New England Journal of Medicine. The senior author is Gluckman. I don't think that in this audience we need to go through the background. This is a comparative study. It is a formal comparison of bone marrow and cord blood transplants in the HLA identical sibling setting, and the main advantage of that is that we can minimize confounding effects of other factors that might influence GVHD risk, particularly differences in HLA compatibility between the bone marrow transplant and the cord blood transplant setting. We focused on children under the age of 15 years, HLA identical sibling donors. We only looked at unmanipulated cord blood and bone marrow grafts. All these transplants were done in the 1990s. Our primary question was is graft versus host disease really less with a cord blood transplant versus a bone marrow transplant in a setting where you have the same degree of donor recipient histocompatibility. We also wanted to look at time to hematopoietic recovery of both neutrophils and platelets of severe GVHD, chronic GVHD and survival. What do we have to consider? And, I show this slide more as a backdrop for our discussions this morning because even when we look at the HLA identical sibling setting we see two groups that have a lot of differences when you look at bone marrow transplant recipients and cord blood transplant recipients, and we have to consider differences like recipient age, sex, weight, CMV status, donor sex, ABO match, conditioning regimen, GVHD prophylaxis, nucleated cell dose which I know that will get a lot of discussion today, and use of hematopoietic growth factors post-transplant. There are a lot of differences in the way transplants are done and in whom they are done when we talk about cord blood transplant recipients and bone marrow transplant recipients, and that has to color our discussions this morning. To try and adjust for these confounding effects we use a statistical approach with Cox, multivariate Cox proportional hazard regression. Now, despite the fact that we used two large international databases and captured what we think was about 90 percent of the HLA identical sibling cord blood transplants that were done in the 1990s, we only had 113. During that period of time, we were able to identify 2052 bone marrow transplant recipients with the same eligibility criteria that had been reported to the IBMTR. Even though we restricted it to children, the cord blood transplant recipients were still younger. There were some differences in gender distribution. They weighed less. There was not a significant difference in malignant versus non-malignant disease and CMV serum status. This is just a list of the diseases for which these transplants were done. There are no surprises here. These are typical transplant indications in children. But, the transplants were done differently. There was a trend toward more use of TBI in the bone marrow transplant cohort, and quite a difference in the type of GVHD prophylaxis used, with many more of the bone marrow transplant recipients receiving combined cyclosporine-methotrexate and many more of the cord blood transplant recipients receiving G or GM-CSF post-transplant in a prophylactic manner to promote engraftment. Nucleated cell dose is about a log different, as one would expect. These patients tended to be transplanted later so, despite trying to select a similar population, they tended to have somewhat more advanced and long-standing disease. We will just cut right to the results. The primary endpoint was Grade II-IV acute GVHD. In both univariate and multivariate analysis there was a significantly lower risk of Grade II-IV acute GVHD in the cord blood transplant recipients, a difference of about ten percent in absolute terms. If we look at the severity of acute GVHD even among those patients who got it, it was less severe in the cord blood transplant cohort. Nine percent of the bone marrow transplant cohort had Grade III or IV acute GVHD versus two percent in the cord blood transplant cohort. Chronic GVHD was also significantly less in the cord blood transplant cohort, less than ten percent versus about twenty percent. A couple of things to notice here -- even in the bone marrow transplant cohort these probabilities are pretty low. Why? These are children. Children do well with transplants whether you are talking about bone marrow transplant or cord blood. Another thing to keep in mind in our discussions is unrelated donor transplants later on this morning. Now the downside -- the hematopoietic recovery, either measured by ANC, as shown on this slide, or platelet recovery was significantly slower in the cord blood transplant cohort. Bottom line -- survival was equivalent. This just shows all patients together. Again, notice the pretty good results. These are children. And, if one separates by malignant and non-malignant disease you see that children who get transplants for HLA identical sibling transplant for non-malignant disease have a very good outcome. Those who get transplants for malignant disease have a somewhat worse outcome. But no difference with either a bone marrow or a cord blood transplant. I promised I would keep this very short because I don't want to get behind right away. So, what does this tell us? It tells us that we have to really consider things other than the graft type when we compare bone marrow and cord blood transplants; that even in the HLA identical setting it seems that graft versus host disease is significantly less with cord blood than bone marrow transplants; that this advantage is offset by a markedly delayed time to hematopoietic recovery. We were unable to look at immune recovery in this cohort; we just didn't have the immunologic data for most of the cord blood transplant recipients. But, the net effect is that the results, as measured by survival, are equivalent with cord blood and bone marrow transplantation. I think I will stop there, and we will be holding all questions until the panel discussion at 11:15. With that background on the HLA identical sibling setting, I will ask Dr. Pablo Rubinstein, from the New York Blood Center, to present the results of unrelated donor cord blood transplants as facilitated by the New York Blood Center. Thank you. [Applause] New York Blood Center DR. RUBINSTEIN: Good morning. It is always a wonderful pleasure to follow Mary and it is a very difficult yardstick to measure up to. The data I will present is abstracted from the clinical results reported to us by the transplant centers which performed transplants of cord blood from the Placenta Blood Program of the New York Blood Center. In summary, there have been 127 patients who received first transplants from our program. These patients have been transplanted since 1993, as you see in this slide. From this slide, there was a very rapid increase in the number of transplants per year, and this increase is becoming slower and then there is a decrease in the last few years for reasons that may be related to the number of alternatives that are now open to the transplant centers as supplies for this material. When we started that was the only option for an unrelated cord blood graft. There are now 50 cord blood banks listed in the BMDW summary. The transplants have been done mostly in the United States but, as you can see here, in a number of other countries, mostly in Europe but also in Australia, New Zealand, Malaysia, the Middle East and South America. The patients transplanted have different ethnicities, as reported in here. Yesterday we went into some depth in the analysis of the ethnicity issue. Just to remind you, in order to provide an appropriate probability of finding a donor it is necessary to manipulate the frequencies of the donors available. That has been done in our program by directing the collections to specific hospitals where the ethnicity of the donors allows us to have proportions approximating those that would yield rather close probabilities of finding matches. You can remember probably that a few years ago Dr. Miti calculated that the probability of finding a match for an African-American patient among African-American donors is only one-third of Caucasian patients from Caucasian donors. Now, as we heard, there was an early demonstration of people doing related cord blood transplants that graft versus host disease was apparently of a lower intensity and lower frequency than with bone marrow. This allowed the concept that it would be possible perhaps to do transplants that are HLA mismatched. In fact, the results of these mismatched transplants through the years have been comparatively good, and many people feel that they actually may be of much lower importance, if of any importance, among cord blood recipients than in the case of bone marrow. The issue of age and size is more complicated since the size of the graft itself is determined accidentally by all kinds of reasons at the time of collection -- that is as much graft as you are going to get. The recipients in our case, surprisingly, belong to all age groups and there are not very major differences in the age groups included, at least numerically. With regard to the diagnoses of these patients, about two-thirds of the patients are leukemia, and among these, about half are acute lymphoblastic leukemia. Genetic conditions altogether amount to somewhere between 20 and 25 percent of the total. In order to do the analysis, therefore, we have defined transplant-related events in our slides that we follow as three conditions that are common to all the transplants -- the possibility of transplant failure by autologous reconstitution, or backup graft, and death. They belong equally to leukemic patients and patients with genetic conditions, while relapse is only applicable to leukemics. So, for this talk we will differentiate between the first three which are common and the last one, separate. Now, the data that we have received is as follows: For myeloid engraftment we have received data from 94 percent of the transplants. Data for 100-day survival is available for 95 percent. One-year survival, for 91; and two-year survival for 93. And, data on graft versus host disease is available on 95 percent of all patients who engrafted. This is a well-known fact by now. The number of patients is bigger but essentially it is the same trend as was shown in our paper in 1998 using about half as many patients. The clear indication here is that the cell dose is a major factor in the speed of engraftment. The slide also includes the medians for the four ranges of cell doses that we have distinguished in these studies -- for 100 million or over, the median engraftment day is 18; for the next group, for 50 to 99 million cells per kilogram, it is 25 days; for those receiving between 25 and 50 million it is 30 days; and for those between 7 and 24 million it is 34 days. So, the influence of the cell dose is absolutely clear, and it is progressive. It is not a threshold event, but it is a steady improvement with the cell dose. Now, for HLA this has been well known, the controversial aspect, and here is the data for engraftment. I should say that all of these slides are constructed by Kaplan-Meier statistics. As you can see here, patients with no mismatch or no mismatch within the definition of our program, which means no mismatch at the serologic level with splits for Class I and the titer solution as available at the time for Class II, shows that those without mismatch engrafted faster and the difference is significant. But there is, surprisingly, no difference between those that differed at one antigen or two. This, in a way, parallels the experience in unrelated bone marrow. The surprising thing that I showed you yesterday, somewhat unexpected, is that when there are blanks so we have an asymmetric mismatch at a single antigen level and the blank is the mismatch, we can have different levels of match in the graft versus host direction and in the rejection direction. Here you can see that while the graft versus host direction, as described in bone marrow, is a little better perhaps than when the antigen is present in both directions, the mismatch, the opposite is not true of the rejection direction. Now, the explanation given for the improvement in the graft versus host direction by the Seattle group is that that reflects the withdrawal of graft versus host activation of the graft and that improves the chances for the graft to overcome whatever resistance may exist in the recipient. The other side, the rejection, was not found by the Seattle group and, in struggling to find the reason for why it occurs here it seems to us that the possibility that needs to be explored is the possibility that there is previous sensitization of the recipient and, therefore, a rejection phenomenon, an immunologic rejection. Now, just as we reported earlier, there is a difference according to prognosis, and there are three diseases that we identified earlier that came out of multivariate analysis as being significantly slower in engraftment. Two of these, severe aplastic anemia and Fanconi's anemia, are perhaps understandable because of the known problems associated with these diagnoses. CML was a little surprising but, as you can see in the slide, the overall engraftment is not all that different from patients transplanted for other diseases but it is a little slower and that effect may be due to the cell dose since the age difference is quite striking. We will see a little more about this later on. One of the factors that was a little distressing to us was finding, as you see in this slide, that the U.S. centers do better than all the non-U.S. centers considered as a single group. Obviously, this is an arbitrary decision from our point of view. It is probably not correct to lump them altogether. There is an enormous variation, but in subsequent analyses there are measurable differences in parameters that we know are important in the overall story. One of them will be seen in this series of slides. The multivariate analysis of the factors that influence the speed of engraftment is shown here. From all of the factors analyzed, and there are a lot of them, these four factors remain significant after Cox regression in more or less exactly the same way as Mary described a little earlier. Now, another factor that is interesting is age. In this slide you can see that the patients where age and weight are independent variables -- this slide refers only to adults and for adults engraftment seems to be independent of age. It is a good sign. If we did this for all patients, of course, there would be an important influence of age because age would be confounded with weight. In younger patients and children age and weight are distinctly correlated variables. Besides engraftment, there are a few other areas that I will explore with you today, and one of them is graft versus host disease. As Mary showed us in the related situation where histocompatibility differences do not exist, there is a clear difference between the frequency and intensity and chance to develop chronicity for graft versus host disease. The same happens for unrelated placental blood grafts. You see here, with the number of mismatches of HLA we increase the number of severe graft versus host disease, in the red bars, or medium to severe, Grade II, in the yellow bars, and we decrease, of course, the frequency of those that have no graft versus host disease. But the overall frequency even for people receiving two antigen and three antigen mismatched grafts is not in the same league as that which follows or has been reported following bone marrow transplants. Now with a larger number of patients and longer evolution, we can tell that there is a significant difference in the probability of having chronic GVHD according to the number of mismatches. Most of this chronic GVHD is of the small variety, so to speak, not extensive chronic GVHD. About survival, the overall data shows, as expected, that there is a better prognosis for patients with genetic disease than for malignancies, and that acquired disease, meaning severe aplastic anemia and similar conditions, are somewhere in between, but these are the overall two-year plots. To remind you, we separated relapse and the total number of patients that suffered failure are shown here -- 23 had autologous reconstitution and 43 required a backup graft. Death from all causes amounted to 424 patients. There were 91 cases of relapse. In general, survival is compared or analyzed in comparison to the cell dose, and we will do that too but we just wanted to show that, as is well known, there is a correlation with the age of the patients. Also as Mary said a few minutes ago, younger patients do better than older patients and this is seen here. Now, for adults the same phenomenon happens. So, this is a little different from engraftment where age didn't seem to make much difference. It does make some difference, but this level is not significant but there is a kind of a trend in which the younger patients seem to do a little better. The numbers are not large. Now, total nucleated cell dose, which is the conventional way to look at these, says that there is the expected correlation. When we look at the curve, it seems that the major difference occurs very early on. The curves are rather parallel from some point and maybe about three months after transplantation they are roughly stabilized. the major differences imposed by the cell dose occur early on. Perhaps that is not a surprise. Now, the age group is, of course, correlated to the total nucleated cell dose, much more so at the beginning, in the early age groups, than later on but these are significant correlations between these two variables. Here is a multivariate analysis attempting to look at both age and cell dose even though they are correlated variables. As you can see, if we study the patients from the point of transplant on, we see that both of the are significant. Age and cell dose have an independent influence but after engraftment the cell dose drops off and age remains the single important variable, or I should say a single significant variable. This is the influence of HLA and I think with these numbers of patients there is little doubt that HLA is important. Of course, I will show you later that this is confirmed by the multivariate analysis where HLA is important when you look at it from the beginning of the graft, and is also important after engraftment. HLA mismatches can be looked in a number of different ways. This is to look at patients with a single mismatch to try to see if one locus is more important than others. We all have the feeling that DR is probably more important but the data do not support a significant difference. Here is the effect of high resolution. It is clear that high resolution is very important. Patients with high resolution mismatches do as bad -- in this slide it looks as though they do worse than those with low resolution mismatches. So, resolution is important for DR. Now, a question that lingers in the mind of many HLA people is whether there are effects associated with haplotype. The patients with one mismatch can have a shared haplotype or maybe they have two haplotypes that are different and only one of them results in a mismatch. We would like to see a difference but this is still not significant. When you have two mismatches, does it make a difference if they are at one locus or at more than one locus? And, the answer is that there is no difference. These numbers are quite significant. The next two slides are a little complicated. Here is when you have two mismatches at A only, at B only -- the first two lines, two loci, or DR only. DR only seems to be different. In this context, however, there is no significance for the overall group but we have to keep an eye out on DR and, in fact, we avoid two DR mismatches like the plague. Here, the same lines are kept but, in addition, we have these lines of combinations -- BDR and AD or ADR, and you can see that there is some difference with the ADR being a little worse but this is not significant. When the two mismatches are Class I or involve one Class I and one Class II antigens, there is no difference. Another factor that is of importance and that we have been looking at is the influence of the experience of the center in the results. This has been seen for bone marrow in a number of patients, reported in the literature, and for cord blood it seems to be the same. Centers with more than 20 transplants are doing better. There are many factors that go into this, not just the experience and dexterity of the transplanters but there is a great deal of learning that is required to improve this overall prognosis, and this is happening and it is not just those with more than 20 but those with 10 that are doing better. To remind you, there is also a difference between the United States and the other centers and, again, most of the difference is at the beginning but we will see these in a multivariate analysis in the next slide. This is the difference between different diseases. Again, CML, Fanconi's anemia, severe aplastic seem to be doing worse but, again, there is the influence of other factors. This is a multivariate analysis. This is from the beginning. As you can see, age, cell dose and HLA match and center experience, U.S. versus non-U.S. and disease are either significant or close. But after engraftment the multivariate analysis only shows two highly significant variables and those are age and HLA match. Just to finish, we have accumulated now 18 patients with a second cord blood transplant. As you can see, this seems to have saved some of these patients. Engraftment is not all that different, and survival for this very small group of patients indicates that at least some of these patients will survive. So, it may be a reasonable option to consider. As you can imagine, there is a very large number of other things. I have some more slides but I think I have exhausted my time and so thank you very much for your attention. [Applause] DR. KURTZBERG: We are saving questions until the panel discussion, and we are going to have a change in sequence because John is stuck on the Metro, I understand. Anyway, he is coming but Donna Wall has kindly agreed to go next and will talk about the St. Louis experience. The St. Louis Experience DR. WALL: Thank you very much. Thank you again for the opportunity to share our experience. Basically, I would look at this talk as building on the groundwork that has been developed by Dr. Rubinstein and basically a validation of his experience from a second cord blood bank. I would also like to take advantage of the time I have to paint a little bit of a picture of cord blood experience at a single pediatric center. I think this might be helpful as policy is being developed in looking at the potential impact of cord blood in stem cell transplant with a pediatric focus -- so, kind of a flip. The St. Louis Cord Blood Bank started in 1996. We explained our basic operations. We have 5500 cord blood units that are available for research and transplant. Our philosophy is that we have a resource for use by centers who are transplanting. So, it has been our approach to make our units as available as possible. In doing so, we have listed our cord blood units with the National Marrow Donor Program, the Bone Marrow Donor Worldwide and Net Cord. In addition, we have recently signed on with the COBLT trial as a transplant center and our cords will be available for use with the COBLT trial for those who do not have matches within the COBLT inventory. To date, we have released 175 units from the bank to 61 centers and 12 countries. From this, 166 transplants have been performed; 22 have really just gone out recently and are not part of this evaluation. We have not received data on 20 of the transplants, and we have 110 that are evaluable. This is fresh experience and these are not going to be as stable curves as Dr. Rubinstein's data. We have not included in the analysis cord bloods that were used for second transplants, cord bloods that were used for expansion. So, you are going to see some of the numbers move around a bit. The diseases being treated are similar to what is being treated with allogeneic transplants, majority of leukemias, and the majority of the leukemias are ALL; immune deficiencies; bone marrow failure; hemoglobinopathies and the metabolic storage disorders. Our median age of patients being transplanted is a pediatric age of 10. We have had 35 adult transplants performed. We have tried to define engraftment/non-engraftment, and the definition that we have used is a neutrophil count by day plus-42 and non-engraftment would be an earlier documented lack of chimerism or death between day 28 and 42, saying that those deaths are more likely due to delayed engraftment. And, I would be very interested in how everyone else is doing this one. Using that definition, 12.5 percent of the transplants failed to reach ANC by day plus-42. When we used day plus-60 as our cut point, we have 7 percent of the grafts, or 6/86 that failed to reach engraftment. We took a look at these 5 patients who engrafted late and they did come in at variable time points and, interestingly, once they did engraft they were all survivors. So, I give you that as information -- what you do with it. Overall, our time to engraftment for the cohort is very similar. Our median time to engraftment is 25 days for an ANC of 500. Similarly, our platelet engraftment -- we are really not seeing much platelet recovery until about day 40 and the median time is about day 60. What we have done is take a look at cell dose impact on engraftment, and when we do that -- and we did this at many cell dose cuts using total nucleated cell dose, and basically at any time point that you look at you will see that the more cells, the better, which is similar to what Dr. Rubinstein has discussed. For us, the biggest cut comes once you are below 3 X 107 cells/kg and that is a recurring theme. Once you are above 3 X 107 cells/kg you are pretty reliably coming in with an absolute neutrophil count before day 25. When we look at survival, we also notice that there is a difference in survival once you are above and below 3 X 107 cells/kg. When we take our cut higher -- and actually a number of pediatric patients are able to make this cut -- we do have really very reasonable engraftment times once we are over 5 X 107 cells/kg or 10 X 107 cells/kg. At this point, our survival curves are really not impacted once we are in this cell dose range. This is actually the lower cell dose per kilogram curve. What we have done is taken a look at seeing whether we can define a threshold for which we say we are above a given cell dose; you will almost assuredly engraft quickly. So, this curve is cell dose per kilogram, and that is 10 X 107/kg and this is days to ANC 500. This is work in progress. This line, here, is basically our median time to platelet engraftment at 25 days. If we take a look at patients who received 5 X 107 cells/kg or, the lower line, 3 X 107 cells/kg, you are seeing pretty much a threshold effect where a majority of the patients are engrafting reliably, and we are sort of fishing for should we be using total nucleated cell counts; should we be using CFU; should we be using CD34 in guiding transplant centers to make a good decision? Similarly, when we look at CD34 X 105/kg for recipients, we find this threshold is a little bit cleaner, where we can define a CD34 cell count above which we are pretty reliably engrafting at a reasonable time period. The same holds true for CFU, which is not rocket science because we know all three correlate fairly well with cell dose. When we take a look at platelet engraftment, it is much more of a scatter plot and there is less of a dependency of platelet engraftment on cell dose. I would like to go back to the original slide looking at neutrophil recovery and focus on those who have less than 3 X 107 cells/kg, and point out that in this population there are those who had this lower cell dose who engrafted very reasonably compared to an equal percentage who were slower to engraft. One of the questions we asked was whether this population had a higher CD34 count. So, could we, on lower cell doses when we are looking for donors for folks, flip over to CD34 count and say this is a rational approach? That we will be able to predict that you will be a more rapid engrafter? In doing so, when we look at absolute neutrophil count less than 25 days/greater than 25 days for those with a lower TNC dose per kilogram -- that bottom half of the curve, we are seeing that the more rapid engrafters had a statistically significant difference, p less than 0.04, but not a really clinically significant difference in cell dose on the CD34 count. Similarly, when we did the flip we took a look at the ones who were rapid engrafters versus slower engrafters who had a low CD34 cell count, and we looked at their total nucleated cells per kilogram and we do not have a statistically significant difference between the two groups. So, we can't flip to the other measure of hematopoiesis in trying to predict at time of choosing a product for transplant, saying that this product is likely to be a better product than not. Overall, survival for all cords from the bank at this point is a week bit better than 50 percent. When we take a look at survival by cell dose, we have the expected difference for cords that were transplanted at less than 3 X 107 cells/kg, but above that we have basically a scatter plot. The highest cell dose is the purple line and then intermediate cell dose. So, basically my read of this is a threshold effect above 3 X 107 cells/kg roughly. We took a look at the same analysis using CD34 cells. We really did not see a difference in long-term survival. You know, our power of the numbers that we have is nowhere near as great as others, but with a sizeable number of transplants we are not getting an absolute sinful number of CD34 upon which not to transplant. Much the same as Pablo, we have a mix of HLA matching that has been accepted by transplant centers. A majority of patients have a 1- or 2-antigen mismatch, and the matching is defined by serologic Class I and high resolution molecular Class II. All typing has been done molecularly. With the power that we have, we are not seeing a difference in HLA matching for survival. In our 6/6 antigen matches compared to 4/6, compared to 5/6, and then a small number of 3/6 matches -- they are not pulling apart at this point, the problem being that we had a very few fully matched and majorly mismatched products. The question is raised as to how big an impact is cord blood going to have on transplantation as an alternative donor source, and that is a question that is going to drive resource allocation -- how many cords need to be banked; what is the expected utilization of cord blood. So, what I would like to do is quickly paint through for the regulators in the audience our experience at Cardinal Glenning Children's Hospital for the last six years as a pediatric transplant program, with an acknowledged bias towards use of cord blood as the alternative stem cell source. During that time period, we have performed close to 100 allogeneic transplants. We are a good Catholic hospital so we have close to a quarter of our transplants being HLA matched siblings, and this is a higher match than most transplant programs will have. We have identified 8 partially matched family members, and the rest are unrelated donor and in that pool, roughly 2:1 use of cord blood compared to unrelated marrow. When we look at overall survival for the program, actually our gold standard is our matched sibling curve, sitting at about 60 percent, and superimposed with the unrelated donor marrow with cord blood and partial matched family members just above. So, basically no major difference in survival between the different allogeneic groups. Out of the cohort of kids that we did treat with cord blood, they ranged in age from 2 weeks to 15 years. All were done on IRB approved protocol. All did not have an unrelated marrow donor available in a timely fashion. They represent very much the major advantages of cord blood. We have a fairly active immune deficiency program, and the metabolic disorders and, if I may just cheat a little bit as a pediatrician can do, this little one had RSV pneumonia, was on a ventilator, in terrible shape; was able to have a cord blood identified and be ready as soon as she came off the ventilator for treatment for underlying immune deficiency, and is a long-term survivor. This little one had a necrotizing pseudomonas pneumonia going into transplant and was transplanted within 2 weeks of diagnosis. This kind of movement and the availability of an alternative donor stem cell source can only be done with cord blood. Similarly, our little ones with leukemia are in a similar clinical state where they have either myelodysplastic syndrome evolving into leukemia, an untreatable disorder with conventional therapy -- very hard to keep these kids in anything kind of condition to get to an unrelated donor marrow transplant. This little one failed autologous transplant. We were able to get her aplastic and then take her into a cord blood transplant. This is my soapbox opportunity, that in developing cord blood utilization strategies, they really have to be flexible and rapidly available to transplant programs. The other major group of disorders that we treat are the bone marrow failures. The majority of the children who are transplanted receive 3-6 antigen matched cord blood units. There were only a few 6/6 antigen matches. Most had major mismatches. Our bias, as a program, has been to try and match for DR although I know the data doesn't support that. The cell dose was in general a healthy cell dose with several of the kids getting well above 1 X 108 nucleated cells/kg. What I have done here is compare engraftment between all our marrow recipients and all our cord recipients. So, this is our matched siblings. This is our unrelated donor marrow and our partially matched family members as the marrow curve, which is in peach. The engraftment time for the patients with cord blood is basically superimposable in our pediatric population. There is what looks to be a little bit of a shoulder of more rapid engraftment with the cord bloods and that is completely attributable to use of steroids in the cord blood and methotrexate in the marrows. But you notice that the time to engraftment is very quick, with a median time in the range of 15 days. Our platelet recovery is slower, as reported by everybody where we compare marrow coming in roughly at 25 days, and the cord bloods coming in significantly later at closer to 60 days. When we look at survival, we are again echoing our larger experience with the cords that we sent out where we are really not picking up a survival difference between any of the groups. The anti-leukemic effect looks like it is real. Of the ten patients transplanted for leukemia, we have had only one relapse. Graft versus host disease, given the degree of HLA mismatching accepted, is well within what we are used to seeing in unrelated donor marrow settings. So, if we look at our overall patients, 18 of the patients had Grade 0-1 graft versus host disease and 7 had Grade 3-4 graft versus host disease and, as Dr. Kurtzberg has mentioned in the past, this tends to be very treatable graft versus host disease. I don't want to under-sell it but it is treatable. Our experience also is that we are seeing very little in the way of long-term chronic graft versus host disease, and the majority of it being limited in severity. So, our conclusions are that cord blood is a very robust alternative stem cell source, especially in pediatrics. With a good cell dose, engraftment time is very similar to that of bone marrow but is delayed compared to platelets, and graft versus host disease does occur. In our hands, we are not able to pick out a difference between the degree of HLA matched but our patient numbers are smaller. I would just like to thank the many people involved in this: my home team, the obstetricians who are involved in the collection program, the Stem Cell Transplant Program, and the National Marrow Donor Program. Thank you. [Applause] DR. HOROWITZ: Thank you Donna. Just to remind you that we are holding questions till the panel discussion. Now Dr. John Wagner, fresh from the Metro, will present the combined Duke and University of Minnesota experience. Thank you, John. Consolidated Data of Duke and Minnesota DR. WAGNER: Sorry for the delay. I finally made it but, unfortunately, I actually missed the first couple of presentations. So, I am at a little bit of a disadvantage but, nonetheless, I am sure we will have a chance in the discussion section to talk about the differences in the analysis that I present versus what some of the others have already presented and perhaps in the future as well. Going back now to old-fashioned slides, what I am going to show you is basically sort of a thought process of why individual transplant centers may have something in addition to offer from the larger registry presentations. Certainly, there are a certain number of advantages in the data that is coming from the umbilical cord blood banks, but there is also something to be said about the individual transplant centers presenting their own transplant data from a variety of points of view. Certainly, the umbilical cord blood banks are sure to have a tremendous potential for evaluating the cellular composition of the umbilical cord blood graft; certainly a variety of infectious disease testing issues, and genetic disease screening issues as well. Obviously, the transplant centers are particularly interested in hematopoietic recovery engraftment, acute and chronic graft versus host disease and opportunistic infection. The umbilical cord blood banks certainly have a large database and certainly there are a number of things that we cannot do at individual transplant centers. As I am sure Dr. Pablo Rubinstein has already shown, there are certain issues, particularly with regard to HLA, that we may not be able to demonstrate because of inadequate patient numbers. On the other hand, there are certain advantages of transplant centers: There is greater treatment homogeneity with a transplant center or group of transplant centers, and there are standardized eligibility and assessment criteria which do not take place in a registry database. Certainly one of our goals, as addressed yesterday, is to come up with information that will help standardize the product in terms of its characterization and also standardization of clinical endpoints, something we will probably want to address a little bit later this afternoon. There are a number of demographic issues that we have already demonstrated in bone marrow transplantation that are key in understanding what will happen to the patient after transplantation, and we don't want to forget that there are certain things that we already know in bone marrow transplantation. Whether or not they hold true for cord blood remains to be determined, but I think that we have to focus at least on those demographics that are clearly important within bone marrow transplant biology. And, all of you who do bone marrow transplants are certainly well aware of these important issues. But there are also aspects to the graft parameters that we also know will be important based on bone marrow transplant literature, and a number of these things we have already addressed this morning, I am sure, and I will address in this presentation as well. But all of these issues are to bear in mind, I think, while we are trying to come up with product standards and we have to keep reminding ourselves of the similarities and differences between the different presentations that these are things that we need to focus on in terms of all of our presentations in trying to come up with these product standards. Now to the data presentation, as many of you have heard before, Joanne Kurtzberg and I have put together a series of patients both at Duke University and University of Minnesota, and this most recent analysis that has never been presented before is now an updated version of 257 patients with 50 eligibility criteria for this analysis. The patients must have had at least 42 days of follow up, and in this case it is actually longer than that. They must have also not had a prior allogeneic transplantation. They had to have a graft that was 0-, 1-, 2- or 3-antigen disparate between donor and recipient. There were several patients that had a 4-antigen mismatch graft that were not included. And, they must have had a preparative therapy. There were several patients that did not receive a preparative therapy for the treatment of immunodeficiency syndromes. So, with that in mind, there were 257 patients that were eligible for this analysis. I should include one other group of patients that was not eligible, and that was those who were enrolled in the COBLT study which I believe was about 24 patients. So, other than those patients, all patients were eligible for this analysis. As you might expect in a primarily pediatric program, about two-thirds of the patients had malignancy, with the majority having acute lymphocytic leukemia but also a variety of diseases, as shown here. About one-third of the patients had a bone marrow failure syndrome, again, pretty much in line with what has been presented before but with larger patient numbers. And, you see there is a high preponderance of patients with a metabolic disorder because of the interest of our type of transplant at these two institutions. With regard to HLA mismatch, clearly the majority of our patients have either a 1- or 2-antigen disparate graft. So, as Donna Wall just presented, clearly, when we talk about the influence of HLA we are really talking about differences between 1- versus 2-antigen mismatched grafts because there were too few patients in the 0- or 3-antigen mismatched group to be discussed in great detail. Nonetheless, we will report the results in all the outcomes for each individual degree of HLA mismatch. When we had 9 that were unresolved, Joanne Kurtzberg and I went through the data and there are a couple of patients for whom we could not determine, for sure, what the degree of HLA antigen disparity was, and we will discuss that in our discussion session if you would like. Recipient serum status is shown here. About 42 percent were CMV positive, again, not particularly surprising because this is a pediatric population. In comparison to what Donna Wall just presented, at her institution where she has indicated that there is a preference to using umbilical cord blood, that was also true at Duke University but was not true at the University of Minnesota. So, the way we select donors is actually somewhat different between the different institutions. Nonetheless, as it turned out, the majority of patients both at Duke and at the University of Minnesota would have been considered high risk on the basis of criteria that are used by most transplant centers. That is, the patients were either in relapse or in third remission, and there were also patients -- which I guess I did not show well here -- who were in accelerated phase CML or patients who had high risk cytogenetic abnormalities perhaps in CR1 or CR2, but they would be the standard high risk patient population for most transplant centers. GVHD prophylaxis was another issue. As many of you know, initially both at Duke and University of Minnesota we both started using cyclosporine and methotrexate which would be considered the standard practice in bone marrow transplant, but because of our concerns about the delay in neutrophil recovery, rightly or wrongly, we very quickly moved on to using cyclosporine plus methylprednisolone. As many of you know, at Duke there was use of high dose methylprednisolone, starting at a higher dose with a slow taper, as compared to the lower dose methylprednisolone initially developed actually at Sloan-Kettering which we adopted and then modified at Duke. Prophylactic G-CSF was not used at the University of Minnesota originally based on the results in sibling transplant using umbilical cord blood. However, based on analyses at Duke, we began using G-CSF at a dose of 5 micrograms/kilogram and at Duke it remained at 10 throughout the study period. We looked at a variety of demographic factors, graft parameter factors and treatment factors that might have potentially influenced the outcome after transplantation. Certainly, our goal was to be able to identify those types of patients that would benefit most by using umbilical cord but, alternatively, those patients who would actually do worse using umbilical cord than might have been expected with a bone marrow transplant. Were there certain graft parameters that would help us predict outcome, or were there certain treatment factors that may or may not be important in the outcome after umbilical cord blood transplantation? At least, that was our initial design of the study. In terms of overall neutrophil recovery, this is not too different than what you have already observed multiple times. The median time to neutrophil recovery is 25 days with this rather large cohort of patients, with a range of 10-59 days. Neutrophil engraftment occurs equally between those with malignant disease versus non-malignant disease. I also should have pointed out in the prior slide that the overall probability of neutrophil engraftment is 92 percent. We looked at the effect of neutrophil recovery by age and, as you might predict because this is related to the cell dose and size of the patient, there is a correlation between age and neutrophil recovery. But, as you can see there on the bottom curve, those patients are over the age of 17 and certainly a high proportion of patients have neutrophil recovery but it is both delayed and ultimately inferior to what would be observed for the very youngest patients, around the ages of 0-1 or up to 2-9. We also looked at the effect of recipient weight. But, nonetheless, all these things really correlate with the overall cell dose, and I apologize because you will not believe the fact that I have actually left out that slide. [Laughter] I can't believe that that has occurred but, nonetheless, that one slide is missing. What I can tell you is that we divided the patients into quartiles and basically what we found is that those patients receiving a cell dose of greater than 1.5 X 107/kg had an overall probability of engraftment of approximately 90 percent. It was those with a dose of less than 1.5 that had the inferior engraftment. So, nonetheless, perhaps we are fine tuning, now as patient numbers are increasing, where that cutoff might be but we can discuss that further. I apologize for that slide not being present. Nonetheless, we also were interested in terms of what is the effect of G-CSF on neutrophil recovery. It would appear in the univariate analysis that G-CSF may be of importance in predicting ultimate recovery. As you can see in the original cohort at Minnesota, not using G-CSF appeared to be associated with delayed recovery. Whether or not differences in ultimate grafting occurred is not clear but, nonetheless, as we have started using G-CSF it appears to be improved but it would appear that there is no clear-cut difference between a dose of 5 versus a dose of 10. As others have already reported, those patients with Fanconi anemia or severe aplastic anemia appear to have a lower probability of ultimate engraftment as compared to the other patients, as shown here. The reason that we pulled out CML individually was that there was some early data that suggested that CML patients might actually be doing worse but, in fact, as we have accrued patients in this population it does not appear to be any different. The numbers are still small. Now, certainly in our own experience we are going to focus on HLA disparity because our results differ somewhat from what has been reported by Dr. Rubinstein. What we see here, at least in terms of looking at neutrophil recovery, is that we find no clear-cut difference between those with a 0-, 1-, 2- or 3-antigen disparate graft but, again, the bulk of our patients are 1- or 2-antigen mismatched and, clearly, between those two there is no difference. This is just one slide and it was meant to show you the correlation between neutrophil engraftment and a nucleated cell dose. You can see here that you might get the impression that there is a correlation. Indeed, there is, however, I wanted to point out to you that when we looked at those specifically who had graft failure, they really were well in the range of those that engrafted and so, therefore, we can't necessarily predict who would have had a graft failure on the basis of their cell dose. I am not sure if the New York Blood Center experience is any different but that was clearly an important outcome that we were looking to find out whether or not we could predict a graft failure and, at least based on cell dose, we cannot. In the multivariate analysis, again, we find no effect of HLA disparity in terms of neutrophil recovery and ultimate engraftment. We do find a significant effect on the base of cell dose. Not surprisingly, CFU-GM also fell out as being a significant factor, but the two are inter-linked with one another. Then, diagnosis was also important. Those patients who had aplastic anemia or Fanconi anemia appeared to have a lower probability of engraftment as compared to those with non-malignant diseases. Then, patients with growth factor appeared to have improved recovery as compared to those that did not. Looking at overall engraftment, 51 percent achieved engraftment by 6 months after transplantation. The one thing I wanted to point out to Dr. Wall, seated back there, you made a comment that it appears to be delayed as compared to bone marrow transplantation. In an analysis that we performed at the University of Minnesota on a case-controlled comparison and also looking at the data from the NMDP there really is no difference between umbilical cord blood transplant versus bone marrow transplantation, though the reasons for the delay may be somewhat different. When you look at age, once again, there is a clear-cut effect of age. Patients who are older tend to have a slower rate of recovery. Again, this is shown here in terms of weight, not surprisingly, and then here in terms of cell dose where you see once again that those patients with the higher cell doses appear to have a better recovery than those with the lower cell doses. The reason why we were interested in looking at the effect of GVHD in terms of platelet recovery is that this is clearly a factor that delays recovery after unrelated marrow transplantation and perhaps with larger numbers of patients we might be able to detect a difference but, so far, those with graft versus host disease appear to have a modest delay in platelet recovery that does not appear to be statistically significant. When you look at the effect of HLA disparity, here you do find some interesting trend in that you find that those with matched or 1-antigen mismatched graft may, indeed, have improved platelet recovery as compared with those with more mismatched grafts -- not quite statistically significant but you are seeing a separation between the two which probably reflects patient numbers and may change over time. So, in the multivariate analysis, again, we cannot detect an effect of HLA mismatch, although there may be something that will develop over time. Again, cell dose appears to be important. Age is important and diagnosis is important in terms of predicting platelet recovery. Acute graft versus host disease is certainly one of the major potential benefits of umbilical cord blood transplantation, not only because we are hoping that there may be reduced alloreactive response but also may allow us to cross HLA barriers. So, even if we do find that there is an effect of HLA disparity which will be important, certainly what it also shows us is that despite HLA disparity so far we have not observed a high rate of acute graft versus host disease. Overall, at this point it remains stable -- the probability of a Grade III or IV GVHD of 12 percent. When you look at the effect of age which would have predicted increased risk of graft versus host disease in the bone marrow transplant setting, thus far we find no difference between the various age groups. When you look at the effect of CD3 cell dose in the bone marrow transplant category, you would have expected a positive correlation between CD3 and acute GVHD, and here you find no such correlation and, in fact, though it is not statistically different, those receiving the lower cell doses actually had the highest graft versus host disease. Clearly, there is no effect of CD3 cell dose thus far. We were concerned that perhaps using a lower dose methylprednisolone might be, indeed, associated with a higher risk of graft versus host disease but, as you can see, all three regimens were virtually overlapping with each other. Grade II-IV acute GVHD was not influenced by HLA match, and I should point out, because of the discussion yesterday, when we are discussing the effect of HLA match with graft versus host disease we are referring to the degree of mismatch only in the GVHD vector. When we are discussing grafting both for platelets or neutrophils we are only talking about HLA mismatch in the engraftment vector. So, we separate those out. This is not just overall mismatch. So, if I showed you the numbers under each degree of mismatch they would differ between the two presentations of engraftment versus GVHD. Chronic graft versus host disease is what remains dramatically low, at least within our own experience. Overall probability of chronic GVHD is only 7 percent. This is dramatically different compared to what we would expect with bone marrow transplantation even with a pediatric age group. When you look at the effect of age here, also considering that maybe we will find a higher rate in the adult patients, thus far we have not. In this analysis there are practically 50 adult patients or at least 50 over the age of 17. When you look at the effect of HLA disparity, you find that there is no clear-cut effect of HLA disparity once again. The one thing that we do find is perhaps a trend towards increased graft versus host disease with a lower dose of methylprednisolone with very little in the other regimens but, again, the patient numbers were small but it does approach statistical significance based on this univariate analysis. Something somewhat interesting that was not expected was that there may be an association with the use of melphalan, at least high dose melphalan. We compared no TBI, that is chemotherapy prep alone, to those that received TBI in combination or without low dose methylprednisolone and TBI plus high dose melphalan. As you can see here, there seems to be an association between the use of high dose melphalan and the development of chronic graft versus host disease. Why that is the case is not clear. When you do a multivariate analysis, again HLA mismatch does not fall out in the multivariate. Cell dose appears to be correlated but I should tell you that the cell dose is actually inverse to what you would have expected. That is, the lower the cell dose, the higher the GVHD. CD3 cell dose, again same thing -- there is inverse relationship of what you would have expected so it is not explainable or, at least, I can't explain it those far. Then, the use of melphalan with a relative risk of 16. When you look at overall survival at one year, we have a survival that exceeds 50 percent, and at four years it approaches 41 percent. When you look at those with malignant disease versus non-malignant disease, it is not particularly surprising that those with a malignancy have a poorer overall survival but, nonetheless, quite respectable considering that the highest proportion of patients had a high risk disease, and also the majority of patients having 2-antigen mismatched grafts. Looking at age, clearly there is an effect of age. Those that are youngest did the best but then, again, those are that are youngest are predominantly with non-malignant diseases. Nonetheless, you see that adults also do respectably well and that will be discussed in greater detail in the next presentation. The one thing that we previously did not observe, we found previously no effect of CMV on outcome but, as you see, as we increase the numbers of patients we are now being able to delineate an effect of CMV seropositivity in a patient having a poor risk in terms of overall survival -- not particularly surprising in view of the bone marrow literature. The one thing that we also wanted to look at which we have never done before was to look at the impact of race. There were 49 patients that were in the minority category, and you can see here it appears in this univariate that minority patients have tended to do worse overall. However, this is a variety of diseases and so it requires looking into in greater depth but, nonetheless, in this first pass we do detect a difference and we never evaluated this previously. Survival is clearly impacted by cell dose. In this lowest quartile you find that the survival is low, and it appears that somewhere in this range of around between 1 and 2.9 there may be an area where we want to cut off our cell dose as being an acceptable graft. Certainly, those with higher cell doses have a better survival. But look at the effect of CD34. I want to point out this is just the University of Minnesota. The reason for segregating the University of Minnesota here is to make two points. One is that CD34 cell dose seems to be quite important in predicting outcome, and there almost appears to be a threshold effect and above 1.5 the outcome appears to be satisfactory; below 1.5 it appears to be very bad. On the other hand, certainly if you combine the University of Minnesota with the Duke University data set, there remains a correlation with CD34 cell dose, but the CD34 cell doses at Duke were higher than at Minnesota which is a reflection, I believe, of the methodologies that were used. So, for this analysis we will probably be forced to separate the presentation into two slides because there are differences in the methods that are used -- an important point in using this parameter as a means of selecting a graft. Graft versus host disease, although it appears clinically not to be dramatically different, it has approached statistical significance and those with GVHD have a slightly poorer outcome as compared to those that did not, and what you find here is that once again we do not detect any clear-cut difference in terms of degree of HLA disparity and overall survival. As you can see, in the green are the results of 2-antigen disparate grafts and you can see in yellow the results with 1-antigen disparate graft. In the multivariate analysis you see, again, that HLA disparity does not fallout; cell dose does; diagnosis does once again. This time, for the first time we are showing CMV seropositivity as having an impact on survival, as does graft versus host disease. I would point out though in this particular analysis race did fall out of the multivariate analysis. Just to touch on the only slide that I have in terms of what was the impact of umbilical cord blood cell dose on those who engrafted, we detect an influence of cell dose in those after engraftment. What this might mean remains to be determined, but it suggests that perhaps cell dose may have an impact above and beyond just engraftment itself -- in terms of immune recovery, or what-have-you, is not clear. But this is something that is worth following, though I did see at the very end of Dr. Rubinstein's presentation that in your analysis I believe only HLA fell out as being important after engraftment, if I saw that part correctly. Now, what does this mean in terms of relapse? I mean, certainly if we find that we have a lower risk of graft versus host disease, does this mean that we have a higher risk of relapse? Though many of you in this room are interested in cord blood, there are many people who do not believe that umbilical cord blood will be of benefit in the long run, particularly because of the concern of the graft versus leukemia effect. It appears that everyone has bought the idea that graft versus host disease is lower frequency after cord blood, but everyone is concerned about the risk of relapse. Certainly, as time goes on this will be declaring itself one way or the other, but despite the fact that 75 percent of malignancy patients would have been in the high risk category, we only have a 20 percent overall probability of relapse within the malignancy patients. So, it is quite respectable thus far. When you look at the individual diseases -- again, certainly greater detail needs to be displayed which is not yet available, but you can see that even in the patients with acute lymphocytic leukemia the probability of relapse has remained low with a median follow-up in excess of two years. Even when you compare standard risk versus high risk no clear-cut differences have yet shown up. We do not yet discern impact by graft versus host disease other than that maybe something may occur with time as numbers increase, and we have not yet determined impact of chronic graft versus host disease. The one thing that has fallen out, and I have shown to you before that we cannot yet explain, is an impact of the use of G-CSF. Not using G-CSF appears to relate to relapse. This has fallen out every time we have done an evaluation and looking at other factors that may have had an impact upon that. But, as you can see here, those that have the higher risk of relapse are those that did not have G-CSF, and you can see that those who had G-CSF tended to have a lower risk of relapse. In the multivariate analysis it was the only factor that fell out as being significant. How this is explained remains to be determined. So, I guess in summary, we believe -- that is, Dr. Kurtzberg and I believe that umbilical cord blood should still be based principally on cell dose when trying to select which graft is the optimal graft. That is also with the caveat that we are talking about choosing grafts that are between 0- versus 2-antigen disparate. The other thing that we have not yet pulled out, which is clearly going to be important, is really whether or not the types of mismatches make a difference. We have actually done a preliminary analysis looking at those with one mismatch and looking at class of mismatch, and we have now detected a difference but, again, the numbers are quite small. Looking at those with a two-antigen mismatch where we have compared those with two Class I versus a Class I/Class II versus two Class II mismatches, again, we do not yet detect a major difference between the three categories. But in those with two Class II mismatches there were very few patients to make any real conclusions. So, clearly we are worried about that group of patients, nonetheless, with our own analysis we have not yet detected a difference. But for the most part, most of the patients have one Class I and one Class II mismatch, and in that situation it appears that we would still want to base it on cell dose as a primary criterion for selection rather than specifically on HLA disparity. What the minimum cell dose will be remains to be determined. I believe it is somewhere above 1, maybe 2. It is going to be somewhere between the two. That remains to be further defined. In terms of what other things we need to be looking at, I think we need to focus on the standardization of CD34 analysis. I think it has an important impact upon graft selection but because different banks and different transplanters use different methods for CD34 analysis, right now I don't think that should be used to choose a specific graft. In terms of banking, certainly it appears from the discussions yesterday that everyone is moving to banking larger grafts. Clearly, more is better in this circumstance and, certainly, we are going to hear more about the expansion potential of umbilical cord blood hematopoietic stem cells this afternoon, and I think the data that we have all presented in terms of cell dose and its impact upon survival is the greatest reason for looking at this ex vivo expansion strategy. Clearly, I have to acknowledge all the work that has been done at Duke University as well as Todd DeFor who has done the statistical analysis. Thank you. [Applause] DR. KURTZBERG: And now the final speaker before the break is Mary Laughlin, who will give us analysis of a pilot adult experience with cord blood transplantation. Adult Experience DR. LAUGHLIN: I appreciate the opportunity to be here this morning to present, as Dr. Kurtzberg mentioned, a focused study on outcomes in adults grafted with umbilical cord blood. The background is, as you have already heard in the past two years, of the use of allogeneic transplant for patients with hematologic malignancies in marrow failure syndromes. The patient description for myself, an internist, will differ from that which has been presented to you of primarily pediatric patient populations. We face the same problems, however, of graft availability given the demographics of American families, and the higher instances of graft versus host disease when we turn to matched unrelated donor or partially mismatched family members as sources of stem cells for these patients. That background sets the stage for this new stem cell source and most of the data thus far has been generated in pediatric recipients, and has included proportions of adults but has not focused simply on that patient population over age 18. These reports have noted delayed hematologic engraftment and reduced graft versus host disease. The focus of our study was to outline cord blood transplant outcomes in 68 adult patients that were transplanted consecutively at five centers during the time period of February, 1995 to September, 1999. Patient eligibility included patients with high risk or recurrent hematologic malignancies. I must emphasize that these are Phase I studies -- patients at very high risk, multiply relapsed; patients transplanted with umbilical cord blood. Second protocol includes patients with severe aplastic anemia, inherited metabolic or immune disorders. Patients must be under age 55, have normal organ function, no available HLA matched sibling donor. Protocols stipulate that they must not have an available unrelated donor by the National Marrow Donor Program, and for many of these patients, their disease status would preclude the time necessary to identify and mobilize a donor. In this study, preferred units were those matched for HLA at 3 of 6 or better, with a minimum cell dose 1 X 107 nucleated cells, whatever that means to the bankers in the audience, recipient weight. No graft manipulation was performed, no ex vivo expansion, no T-depletion other than hetastarch red cell depletion. Preparative regimens included TBI-based or busulfan-based and serotherapy was provided to all patients, ATG at dose 30 mg/kg, day minus 3, 2, 1 prior to infusion of the cord blood unit on day zero. Graft versus host disease prophylaxis varied among centers but included cyclosporine and the variable was the dose of steroids. Steroids were tapered generally by ten weeks post-transplant and cyclosporine was tapered anywhere between 6 and 9 months post transplant, depending upon whether the patient was exhibiting symptoms of graft versus host disease. Supportive care included use of G-CSF. The dose of G-CSF varied among centers between 5-10 mcg/kg/day, starting day 0 and continuing until full neutrophil recovery. None of the patients were supported without growth factor. Our statistical analysis was a retrospective analysis with follow up as of August 2. The endpoints in this study included kinetics of hematopoietic recovery, event-free and overall survival. Univariate analysis was performed using log rank or Wilcoxon. Multivariate analysis incorporated the Cox proportional hazard regression model. The variables that we studied included age of the patient, gender, HLA and ABO matching, weight, disease entities, CMV serology status of the recipient pre-transplant, and graft characteristics. A few definitions -- myeloid recovery was defined as the first of three consecutive days after transplant when the absolute neutrophil count remained above 500 per mcl. Platelet and red cell recovery were defined as the first of seven days when the platelet count remained above 20,000 and the hemoglobin level remained above 8 g/dl without transfusion support. Chimerism was evaluated in all patients using one of three techniques -- in situ hybridization for patients with sex mismatched grafts; DRB1 in cases in which the donor and patient differed in HLA DR; and use of quantitative PCR for microsatellite DNA markers. In this study, primary graft failure is defined as absence of donor-derived myeloid engraftment by day plus-42. Further definitions -- overall survival was measured from the date of transplantation to the date of death, and censored at the day of last follow up for survivors. Event-free survival was measured from the date of transplant to the date of relapse or death, whichever occurred first, and was censored at autologous hematopoietic recovery or date of last follow up, whichever occurred first. With those definitions, grafts were characterized for HLA using serologic typing for Class I and high resolution DRB1. In only two patients was a 6 of 6 matched graft infused; 18 patients received a 5 of 6 matched graft; 37 patients received a 4 of 6 matched graft. And, 48 of these patients, 71 percent, received cord blood grafts that were disparate at two or more HLA antigens and half the patients approximately received Class II mismatched grafts. The median weight in this series was 69 kg and ranged from 41 kg to 116 kg. The median age in this series was 31 years, and patients ranged in age from 18 to 58 years. What I think is unique about this data set is that as opposed to some of the reports from pediatric programs where a few patients with pediatric diseases may live into their 20s, this patient population does represent the typical adult patient population that an internist would see and refer to an oncologist. And, 50 of the 68 patients were considered high risk, as I mentioned, using IBM-TR criteria; 13 of the patients were non-malignant; 7 of the patients had failed prior transplants; 15 patients had ALL; 19 patients were diagnosed with AML; 15 patients were diagnosed with CML, of which the majority of the patients were either in accelerated phase or had had blast crisis and had received induction chemotherapy to reattain chronic phase. One patient was diagnosed with CMMl; 2 patients had refractory recurrent Hodgkin's disease; 1 patient had lymphoma; and 1 patient had CLL. In the patients with non-malignant disorders, one patient was diagnosed with Blackfan and Diamond anemia and had been transfused for a period of 20 years. One patient was diagnosed with ALD. There were four Fanconi's patients in this series; two patients with myelodysplasia that had not demonstrated evolution to acute leukemia; four patients with severe aplastic anemia; and two patients with myelofibrosis. The graft characteristics in this series -- cryopreserved cell dose, the median cell dose was 2.1 107 per kilogram and ranged from 1 to 6.3. Infused cell dose was 1.6 107 per kilogram and ranged from 0.6 to 4. CFU was infused and CD34. The take-home point of this slide is that these numbers are fully a log less than that which we would normally infuse using a traditional allogeneic graft. CD3 infused dose was similar to what we would see with traditional bone marrow grafts. Here we have plotted cryopreserved cell dose versus infused CD34 and with superimposed regression, the correlation coefficient here of 0.5. I think this is an important graft to emphasize because for the clinician many times this is the piece of information that they have with the HLA typing of the graft that allows them the opportunity to make the decision whether to proceed in treating their patient. I think it is reassuring that this 34 surrogate analysis correlates well with cryopreserved cell dose. Further graft characteristics that we analyzed -- CD34 reinfused versus CFU, again, correlated with one another and, finally, cell dose cryopreserved with infused CFU. The kinetics of hematopoietic recovery in these patients receiving non-expanded grafts median day to ANC greater than 500 was 27, and ranged as early as day 13 and as late as day 59. The actuarial probability of neutrophil recovery in this series was 92 percent, with 95 percent confidence interval as outlined here. Median day to platelets greater than 20,000 was 58. Median day to platelets 50,000 and 100,000 is as outlined. In our MUD transplant patients at our shop, these numbers are also similar as far as kinetics of platelet recovery, again, likely for different reasons. Median day to hemoglobin greater than 8 g/dl was 60 days. In all patients who engrafted there was complete donor chimerism, and there had been no late graft failures observed. A Kaplan-Meier curve of day ANC 500 with 95 percent confidence interval, median here day 27 and ranging from day 15 to day 59. When we broke out the Kaplan-Meier curve kinetics of neutrophil recovery versus the cell dose cryopreserved, breaking it here at the median, we saw faster kinetics of neutrophil recovery in those patients receiving the higher cell dose. Also, when we broke out infused CFUs into these groups, as outlined, and analyzed the Kaplan-Meier kinetics of neutrophil recovery, again, we saw faster neutrophil recovery in those patients receiving a higher CFU cell dose. This is the Kaplan-Meier plot of day to platelet independence -- I apologize for the misspelling. We did not analyze, either by univariate or multivariate analysis, kinetics of platelet recovery because in this series that comprised 32 evaluable patients and we are awaiting further numbers of patients in this analysis. Outcomes in this series -- there were 8 early deaths prior to day 28 and those patients were censored. In the remaining 60 patients, 5 patients demonstrated graft failure. The day 100 survival in this series is 50 percent. This is very high I think and is reflective of two important issues -- patient selection and delayed hematopoietic engraftment. Grade II-IV acute graft versus host disease, despite infusion of grafts mismatched at more than one HLA loci, was 60 percent, with the confidence interval as outlined here, and the incidence of Grade III-IV graft versus host disease in this series was 20 percent, which is representative of 11 patients. In our series of adult patients the incidence of chronic graft versus host disease, 12 of 33 evaluable patients, was 38 percent. This was limited in all patients and non-progressive, except one patient. In our analysis there was no association between graft failure and infusion of HLA disparate grafts, infusion of Class II disparate grafts, CMV serum positivity in the recipients, whether the patient was diagnosed with a malignancy versus a non-malignant disorder, and breaking out their diagnoses. There was no association between the grade of acute graft versus host disease and the graft HLA disparity, whether or not the graft was disparate at Class II with the recipient, CMV serum positivity in the recipient or whether the patient was conditioned with a TBI-based versus chemotherapy-based regimen. Further outcomes in this series, the median follow up of our survivors is 22 months, and ranges from 11 months to 51 months. We have observed four relapses. There seems to be maintained graft versus leukemia factor in this series of very high risk patients. Three patients were diagnosed with ALL and relapsed with their disease. One Hodgkin's disease patient relapsed at these time points post-transplant. Event-free survival at four years is 26 percent, with confidence interval, and this represents 18 patients; the Kaplan-Meier of event-free survival with the confidence interval as indicated, with follow up out to four years. The Kaplan-Meier here of the risk of relapse in this series is 16 percent. When we analyzed event-free survival by CD34 infused there was improved event-free survival in those patients receiving a higher than median dose. Also, when we analyzed cryopreserved cell dose versus survival, there was a trend towards improved survival in those patients receiving the higher cell dose, however, this did not attain statistical significance. There was no impact of receiving an infused higher cell dose versus lower cell dose on event-free survival. When we analyzed event-free survival versus age in this paper, an analysis focused on adult patients alone, we found no significant difference in event-free survival when comparing patients less than age 25, ranging in age 25 to 40 versus those patients over age 40. There was a trend towards less improved survival, however, not statistically significant. We further analyzed event-free survival by HLA disparity and we did collapse this data to combine those patients who received either a 5- of 6-matched graft or a 6 of 6 matched graft to attain equivalent numbers of patients in each study group. There was no significant difference when you compared HLA disparity among the groups. There was a trend towards an inferior outcome in those patients receiving a 3 of 6 matched graft. Somewhat surprising and I don't want to overstate it -- this is 69 patients, but here you would expect an improved event-free survival in patients with 5 of 6 or 6 of 6, but this curve is actually those patients receiving a less well matched graft -- again, a trend and not statistically significant. When we went on to analyze whether this observation was attributable to disparity at the Class I loci or Class II loci, again, we observed a trend of improved event-free survival if you received a graft that was mismatched at Class I rather than better matched at Class I. I don't wish to overstate this observation, this is 68 patients and it is not statistically significant. It may represent the selection by the transplant clinician who, many times, will choose a less well matched graft of higher cell dose, over a better matched graft of a lower cell dose. When we broke out event-free survival versus patient disease entities, comparing patients with ALL, AML and CML versus others, there was a trend towards improved survival in the CML patients compared to ALL patients but this did not attain statistical significance in the univariate analysis. This differs from observations in larger data sets. My only explanation is that this cohort of CML patients had perhaps different characteristics. They tended to be patients in accelerated phase or blast crisis and, therefore, a good proportion of them had received chemotherapy at a time period prior to transplantation. I don't know but that may have had an effect. When we analyzed event-free survival versus CMV in the recipient, here negative versus positive, we did see a trend towards improved survival but, again, this difference was not statistically significant in this series of 68 patients. Further univariate analysis included whether comparing 6-month survival and using log rank, whether the unit was hetastarched, whether the patient received a busulfan-based versus a TBI-based regimen, whether the patient was diagnosed with malignancy versus non-malignant condition, and none of these variables were statistically significant. Further analysis -- whether the graft was matched at Class II, whether the patient developed chronic graft versus host disease -- the caveat here though is that this was a small number of patients, 36 of these patients were non-evaluable due to death prior to day 100. Whether they were high risk versus low risk using IBM-TR criteria, there was no significant difference using log rank analysis. The multivariate analysis in this series, comparing event-free survival, included CD34. We used relative risk 0.8 associated with an improved event-free survival. The observations that we had made in the univariate analysis emerged in the multivariate analysis with a relative risk of 2.5. The diagnosis of CML compared to patients with ALL trended towards an improved survival but was not statistically significant. Female patients fared worse compared to their male counterparts and advanced age was associated with poorer outcome in event-free survival in the multivariate analysis. So in summary, cord blood can be successfully engrafted into adults with refractory hematologic malignancies and marrow failure syndromes. Despite a high level of HLA disparity, cord blood from unrelated donors is associated with a low incidence of a severe acute and chronic GVHD in these adult recipients. There is delayed time to hematopoietic recovery observed. I guess like my pediatric colleagues, I do want to take at least one step out of the statistics to describe a number of the statistic. This person, Chris, is 28 years old, likes to ski in Colorado, was diagnosed with Hodgkin's disease, and attained complete remission status, however, presented to our transplant program with thy-related AML 14 months later. For this individual, no sibling match was identified and the pace of his disease precluded the necessary time to identify and mobilize a MUD donor. I think as we learn more about this new stem cell source, we can be assured that even in adult recipients a proportion of patients can derive direct benefit from use of this allogeneic stem cell source. This patient is now two and a half years out from his cord blood transplant. I would like to also credit my collaborators, Pablo Rubinstein and Cladd Stevens, Joanne Kurtzberg and David Rosari of Duke University, John Wagner and Dr. Barker at the University of Minnesota and Mitch Carro at Lombardi, and my colleagues Hillary Lazarus, Stan Gerson and Omar Koc at Case Western Reserve University. Statistical analysis in this study was performed by Pingfu Fu of our epidemiology and biostats department. Thanks. [Applause] DR. HOROWITZ: Thank you very much. Amazingly, we are ahead of schedule. So, we will take our break now before our final presentation and we would ask you to come back a half hour from now, at 10:30, and hopefully we are going to have a little extra time for discussion. [Brief recess] DR. KURTZBERG: I would like to invite everybody to come on back in so that we can get started again and keep our advantage for being ahead of time so we have time for discussion. It is my pleasure to introduce Dr. Takahashi from Japan to tell us about the experience with the Japanese cord blood banking and registry. Japanese Experience DR. TAKAHASHI: Thank you, Dr. Kurtzberg. I would like to thank the organizers of this workshop for giving us the opportunity to report our experience in Japan, and I would like to talk about the kind of situation of cord blood banking and transplantation first, and then to report the clinical outcome. In our country, about 1500 patients need allogeneic hematopoietic stem cell transplantation every year. The number of bone marrow transplantations has been increased but we can see that they reached a plateau in recent years. On the other hand, shown in green and red, particularly the cord blood transplantation increased very rapidly. We don't have the statistics of 1999 but half of the hematopoietic stem cell transplantation in children are done by cord blood transplantation. This is a brief history of cord blood transplantation and banking in Japan. The first related cord blood transplantation was done in 1994, and the first unrelated cord blood transplantation was done at the Yokohama City Hospital, University Hospital, in 1997. Since April, 1998 the cord blood transplantation was covered by the national health Insurance which was approved in 1998. So, that was very good news for the patients. Now, 270 unrelated cord blood transplantations were done until the end of March of this year. The first cord blood bank was set at the Kanagawa Prefecture in 1995. At that time the concern about the safety of cord blood transplantation and then the quality of the cryopreserved unit. The government research group made a standard for the cord blood banking and also the indication for cord blood transplantation in 1997. Also, in the history in Japan was that we have a very good group which supported to establish the National Cord Blood Bank in Japan. They are so active that because of that Health Ministry got a lot of pressure by the Japanese people and they assembled the Cord Blood Transplantation Study Group in 1997, and we are the technical members for this group and we established clinical guidelines for cord blood transplantation in 1998 based on this preliminary standard. The guidelines are almost similar to NETCORD, COBLT or New York Blood Center standards. Finally, Japan Cord Blood Bank Network was established last year, last September. Now in the Japan Cord Blood Bank Network there are 3000 cord blood units cryopreserved in nine local cord blood banks. The objectives of this network are to promote unrelated cord blood transplantation patients in Japan through collaborative management and joint ownership of information. The establishment was on August 11, 1999. The financial support comes from the Ministry of Health and Welfare, but the financial support is not enough to cover all the expenses. The support covers a part of HLA testing and they also give us the equipment, such as freezer, cell counter and so on, but still the support is not enough. But the basic idea is that the government supports the banks but there should be a competition among the local banks to serve better the patients and hospitals and also the quality. So, there should be competition among the banks, and also every bank should get their own financial support from other places. The main office is in the Japanese Red Cross in Tokyo, and then there are two committees, the administration committee and evaluation committee, and there are main working parties, such as internal system working group, infectious control, distribution of transplant centers. Now, the big problem for us is that there is no price for the unit which was transplanted. That means the units are given to the transplant centers free. So, we have to solve this problem as soon as possible. The administration of cord blood banks -- there are nine cord blood banks registered after government inspection. The projects of the Japan Cord Blood Network are the standardization of collection, separation testing, cryopreservation shipping, transplanting etc., evaluation of management of local cord blood banks and quality control of cryopreserved cord blood units, collection of 20,000 cord blood units in five years, the construction of Internet access system for cord blood search, international collaboration in banking and transplantation of cord blood, and the registration of transplant centers. This is how we get to the national goal of 20,000 units. This is based on Prof. Juji's and Akaza's report. They analyzed HLA type among Japanese, and you can see this is the donor size and this is compatibility, and this shows 6-antigen mismatch serologically and genotypical. Here is 1-antigen mismatch. If the cord blood transplantation is acceptable in 1-antigen mismatch, we can say here if we have 20,000 units of cord blood 19 percent of the patients should be covered, could find their matched unit. So, therefore, 20,000 units is our first national goal. The government inspected nine local cord blood banks and they have made an inspection team and we helped them and we made a checklist, and two of the banks were approved without any change, and others were asked to include several points, such as improving the facility and better document management and centralized processing center. There are nine banks but there are fourteen processing centers in Japan, but several processing centers are now centralized. I would like to introduce our bank as an example of Japanese local cord blood banks. We started collecting cord blood in our department in 1996 following New York Blood Center's protocol. Then we joined the NETCORD. We are the founding members of the NETCORD. The first cord blood transplantation was done to a 28-year old woman and that was the first success patient in Japan. In 1999, in September, we joined the Japanese NETCORD and they asked us to collect 800 units last year and 1200 units this year, and this is the largest number in the network. I would like to report briefly about this patient later, and now we have 1667 cryopreserved units, transplanted to 47 patients, 17 adults and 30 children. Our institute focused on transplantation of the cord blood to adults. So, we have the experience of 20 in our institute and 17 others and 3 are children. This is our process room. We have a clean room P-2 level and the cleanness of the air is NASA 10,000. The clean room is divided into three booths and each booth has a hipofilter and purity is going up to 1000, and the benches also, of course, are just 100. We use bio types. In Japan all the banks use the Hiss method now and two of the banks and three processing centers use bioarchive. I want to just briefly introduce our experience of this patient. There was a nuclear accident in Tokai, near Tokyo, last year on September 30th. The three workers at the uranium convergent front were exposed to radiation where they were mixing too much uranium with nitric acid. This is a picture of how our patient poured into the tank. This is completely out of the SOP. [Laughter] The three workers were exposed to the radiation and a 35-year old man was in serious condition and he received stem cell transplantation. The 30-year old man was in less serious condition. He received cord blood and he didn't have any HLA matched sibling, unfortunately, but we decided to do the cord blood transplantation for him. We found one unit in a bank and this is the patient and this is the donor of cord blood based on the DNA typing, 1-antigen mismatch, and 2 X 107 nuclear cells were there and as preconditioning ATG was administered for two days. Then GVHD prevention was cyclosporine-A and growth factor was administered to the patient expecting rapid recovery of neutrophils. This summarizes all the treatment. You can see the rapid drop of the hematological cells, and on the day of the transplantation, on day 16, the neutrophils came out and reticulocytes on day 22, and the platelets on day 26. The patient survived for 7 months but, unfortunately, he passed away last May because of severe burn of the radiation and at the last multiple organ failure, he passed away. So, I think this is a hard case for the Japanese people but I think we can find a good advantage of cord blood banking on transplantation which used our system in such an emergency situation. Now, back to the Network, this is our system. We have a main server, two of them. One is in Tokyo and one we put in Kokkaido. We are afraid of an earthquake. Each of the nine banks send cord blood information to the main server every week, like HLA typing, all the family information to the main server. And the patient and physician can search through this main server. Everyone can see if they know their own HLA typing, and if they find the matched unit -- 1 mismatched and 2 mismatched antigens are mismatched unit in the order of their dose. After they find their matched unit they now contact with transplant centers but now they have the ID number and the password. They search the second one and then they can find out more deep information about the unit. After they decide to use the unit, they talk to the local cord blood bank because the bank's name of the unit is put in the computer so they can find it. Then they send by fax or they telephone and decide on the final decision. After the decision, they keep the unit for three months and then the local bank submits to the main server the results. Only one patient just has the right to reserve only one unit. All the banks send all the data to the main server. We just have a minor program. If we are fortunate, we can start the search within that week. We are also preparing our English version, and we hope that foreign country people can access the Network. This is the first page of the website, linked to other banks also, and this is the screen on which the patient can type the patient HLA and the body weight and other things. Now the clinical outcome -- the clinical data are summarized by the Basic Clinical Study Research Group Cord Blood Transplantation, chaired by Prof. Saito. These are members of the research group, but there is a strong argument on who should collect all the clinical data -- should it be the bank, or should it be transplant centers? And, there is a strong opinion that academic societies, such as Hematopoietic Stem Cell Transplantation or Tokai, such a neutral society should collect the data and analyze it. So, our information is still limited but we have all the data sent to the research group. Until the end of this March, 207 transplantations were done for 203 patients. There were 121 males and 86 females. The age was 0 to 49 and the median was 6 years old. Body weight, from 5.3 to 69, the median was 18 kg. The total transplanted nuclear cells are 0.62 or 16.6 X 107/kg and the median was 3.4, 7 X 107. The indication of disease categories of unrelated cord blood transplantation -- acute leukemia, immunodysplastic syndrome, lymphoma, solid tumor, bone marrow failure, congenital immunodeficiencies, etc., and metabolic disorders, and the nuclear accident. So, more than half of them were leukemia patients. Engraftment -- the number was 170. Recognize, myeloid engraftment was in 136 patients. The rate is 80 percent. Autologous reconstitution was observed in 9 patients, 5.3 percent. The transplanted total nuclear cell dose and engraftment -- there is no strong association in these two factors. HLA disparity -- more than 70 percent were antigen mismatched and 10 percent were true match and 25 percent were 2 anti-mismatched. The severity of acute GVHD -- the variable number was 145 Grade 0, or 32 percent; 1, 39 percent; 2, 24 percent; and Grade III and Grade IV, 3.4 percent. Moderate acute GVHD total was 39.2. Severe GVHD was 50.1 percent. There is no association between HLA disparity and severity of acute GVHD, as shown in this slide. You can see that 85 percent of the patients showed no chronic GVHD; 5 percent were positive; 8 percent as indicated here; and extensive case was 2 percent. This is Kaplan-Meier univariate analysis that we have used in the study, and overall survival was about 58 percent, and the follow-up median is 158 days. This is disease-free survival, shown here, and it is about 50 percent, and engrafted patients are shown here. HLA disparity and clinical outcome, disease-free survival. There are two haplotype identical transplantations that were done in our institute. This is those two, and we did data analysis by log rank test and the p value was 0.5 but we see good survival here, but there is no statistical difference yet. This is nucleated cell dose disease-free survival. It is complicated. So, we divided into higher than 4 X 107 and lower than 4 X 107. You can see the clear statistical significance here. Age -- if we divided the age to younger than 3 and -- sorry, 16 years old and the others, we see the statistical difference. Yes, age is also an important factor for clinical outcome. And, disease categories and disease-free survival -- we divided this into malignant disorders, congenital disorders, MPD, MDS, SAA, and there is no difference. This is still on colleague outcomes among the acute leukemia patients, the total of acute leukemia, and you can see that it worse on CR1 and CR2. But for AML patients we still couldn't see the difference, but in ALL patients we can see the status on CR1 and CR2. The cause of deaths, 23 of them, in 79 patients MOF, VOD, GVHD, relapse others and unknown cases. So in summary, we have established the Japan Cord Blood Bank Network and the computer search system will be operational this month. The patient and physician can search the patient's matched unit through the Internet and the decision to use the best unit for transplantation will be made by the transplant center physicians, not by the banks. In summary about the clinical outcome, it is still early to analyze the data more statistically. We have only 207 cases. The following factors may be considered as a favorable disease-free survival in cord blood transplantation a total nucleated cell dose, age and disease status and HLA match. This is my last slide and I would like to thank the many American scientists, physicians who helped us a lot from the beginning when we started, such as Pablo Rubinstein, Dr. Carro, Joanne Kurtzberg -- so many people, Wagner and Cladd Stevens, of course, and then the help to establish guidelines, and because of that help I think we could progress our cord blood banking and transplantation much faster than we expected. Thank you very much. [Applause] DR. HOROWITZ: Thank you, Dr. Takahashi. I would like to ask all the previous speakers to please come up for the panel discussion. I know there are a number of people who have had some questions that they wanted to ask in the morning, so why don't we start this session with just opening up the discussion with questions from the audience? Discussion: Can we Define an Acceptable Unit for Transplant? DR. KURTZBERG: Please remember to identify yourself. DR. LANE: Yes, Lane, San Diego. Actually, I am going to start out the questions by asking a question we talked about yesterday, and that is a number of speakers today talked about cell dose once again, and specifically CD34 dose. John Wagner showed a very nice correlation between CD34 dose and engraftment, and Mary Laughlin, I thought, had just some extraordinary data relating CD34 dose to, I believe, not only engraftment but overall survival. So, what will it take us then to move from total nucleated cell count to using a CD34 dose as at least a very important measure of evaluating these grafts? I showed you data yesterday that at least within the Red Cross system of cord blood banks the variability in CD34 dose is really not very high, not much more than one would expect. So, how can we get there? DR. KURTZBERG: I would like to take a stab at answering that. I think we have to standardize methodology between centers to be able to answer that question because even the Duke-Minnesota databases are a good example. We have relationships in both databases but the absolute number is different. And, I think until the methodology is standardized or we have some key that sorts out how, you know, Japan correlates with New York, correlates with COBLT, correlates with San Louis, etc., we are not going to know how to use the numbers themselves. DR. WAGNER: So, just to follow up on that, I mean, basically for both Duke and Minnesota, both individual institutes showed a correlation but the absolute numbers were different so, therefore, I came up with a slide that specifically stated -- or maybe a statement in there that the middle number should be 1.5 X 105/kg. That may be true for Minnesota but it may not be true for anyone else because of the way we do the analysis. So, it has tremendous implications for people in the way they choose a graft. Until we have standardization we can't use CD34. DR. KURTZBERG: The other thing to mention, and I think this is true of Minnesota too, but all of our data is post-thaw, and although we say we have a method that correlates pre-cryo and post-thaw, it is only going to be in the COBLT study that we really have a prospective to look at that. DR. MCNEICE: Joanne, may I make a comment on that? Ian McNeice, from Denver. I am not as convinced from the data I have seen that what people are calling correlations certainly aren't convincing to me when you have an ANC grade of 0.5. I would like to see it be a lot more rigorous before we make that sort of comment. I am still not convinced the data suggest that there is a good correlation that you can predict from the 34 number what is going to happen with engraftment. DR. WAGNER: Well, just to follow up though, there is a suggestion by at least one analysis to suggest that there may be a threshold. So, there may not be a correlation that is continuous, but it may be that it does provide us with a threshold dose beyond which we find no real difference between CD34 doses but there was a lower limit. DR. MCNEICE: And, I would question how rigorous the analysis can be for those numbers. Then when you look at the actual grafts, the majority of the grafts run between 2-4 million total 34 cells -- DR. WAGNER: Right. DR. MCNEICE: So, there is not a lot of play to actually say this is a much better graft than this one. So, I think we need to be careful before we make such a bit leap. DR. WAGNER: We are almost ready to identify a number below which it is an unacceptable graft. That is the goal. DR. MCNEICE: Well, I still don't think that we have the data that supports that that is a true indicator of the engraftment potential of that product is. DR. KURTZBERG: Let Pablo make a comment. DR. RUBINSTEIN: It is just to alert you to the fact that this afternoon there will be some discussion on the hematopoietic stem cells and precursor cells. So, part of this answer may be clarified. From our own data, it is clear that the colony-forming assays predict a little bit better -- not much better but a little bit better than the total nucleated cell dose, and the CD34 count done by Dr. Fisher, with some modifications from the standard procedures, do correlate extremely well at the level or r-squared of 0.88 or 0.9 -- I don't remember exactly -- with the colony-forming cell dose. So, basically, the total nucleated cell dose remains largely enough to account for the quality from the point of view of these two other parameters. They are better, but not very much better. DR. KURTZBERG: Cladd? DR. STEVENS: I would just like to make a comment to help clarify what seems like an apparent disagreement between the Duke-Minnesota data and some of the data that Pablo showed earlier about effective cell dose post-engraftment on survival. If we do a univariate analysis on the total blood center data, you also see a correlation with cell dose. I think the point that Pablo was trying to make was that when you do it in a multivariate analysis the effective cell dose drops out and the only thing remaining is the effect of age. So, it is clear that we have a situation here where we have confounding between age and cell dose, and I think those are issues that need to be considered in a little more detail. DR. KURTZBERG: Carolyn? DR. HURLEY: Carolyn Hurley, from Georgetown University. I, of course, wanted to ask the question about the role of HLA which seems to still be unclear, and I wanted to ask the panel about what their plans were in terms of using high resolution methods to look at matching because, obviously, underneath the issues of are things matched to the serologic level there is extensive disparity that is not being detected. So, that is one question. The second question is the issue of whether transplant centers or banks are making some planned collection of patient and donor samples in order to go back or be able to go back and address these issues of histocompatibility? DR. RUBINSTEIN: Thank you very much. That I think is a very important question. There is further heterogeneity that we have been able to analyze so far. Class II is fairly okay, but for Class I all we know is about the level of resolution equivalent to serology with all splits. There will be additional information coming out from that. It is a retrospective analysis. We do have samples on all of the units and most of the patients, but not all of the patients. So, that retrospective analysis is being done to the extent possible as we speak. DR. WAGNER: Just to complement what was just said, you know, we also agree that it is one of the things that needs to be done and we, in the institution, don't have a large enough series to be able to answer any of those questions. However, we have samples on all the patients but not all the donors. DR. KURTZBERG: In the COBLT studies there are samples on both. DR. WAGNER: Just a comment on trying to determine the role of HLA versus cell dose. I mean, I know that you are well aware that even in the bone marrow transplant setting where we have thousands of donor-recipient pairs to analyze, it is very difficult to separate out the independent factor of specific HLA mismatches, number of HLA mismatches in a setting where you have so many confounding variables affecting transplant outcome. In this setting we don't have thousands of donor-recipient pairs. We have inadequate typing on many of those donor-recipient pairs, and we have confounding cell doses, and an issue in trying to separate out whether it is cell dose versus HLA disparity is, I think, confounded by some of the selection pressures on whether you do a cord blood transplant that is mismatched. If you have a patient with very poor risk, very poor prognosis with conventional therapy, you are more likely to take what you might consider a marginal graft. So, although we would like to be able to definitively say this is the contribution of HLA and this is the contribution of cell dose, I don't think the data exist right now. I think the most that you can say is cell dose is important; HLA is important. The relative contributions remain to be determined. DR. KURTZBERG: I guess the one thing I would say is that I think this is much more complicated than just engraftment and you are okay versus engraftment immune reconstitution and a bigger picture. If you look at everybody's curves, most of the deaths are in the first 100 days and at least half of those deaths are infections. I know in my experience there are equal numbers of infections occurring in engrafted patients, meaning patients who achieved an ANC of 500, as there are in people with delayed engraftment. So, there is not the same protection from engraftment that I think we are all used to seeing with the bone marrow transplant patients. My personal theory is that some of this may be related to immune reconstitution which may be more delayed in the HLA mismatched setting, or may have some other factor contributing like age and capacity for immune reconstitution. But, there is something else driving the fact that these people are getting infections, and that is their main reason for failure. Maybe it is the cell dose threshold; maybe it is immune reconstitution; maybe it is HLA. Probably it is all of the above. But if we could sort that out this would be a lot more successful as a transplant source, and I think it is most important to sort that out for adults because there the cell dose options are less generous. DR. BROXMEYER: Hal Broxmeyer, from Indiana University. Two short questions: How many cord blood transplants have actually been done? Because that is not clear from anything I have heard or read. Are we in the 2000 range? Does anybody really have a handle to be able to at least estimate how many cord blood transplants have been done? DR. KURTZBERG: We have done 290 unrelated. DR. BROXMEYER: That is not helping me. DR. HOROWITZ: We are in the 2000 range. DR. BROXMEYER: You think so? DR. HOROWITZ: Yes. DR. BROXMEYER: Good. It is not that easy to get that answer. The second question deals with time to neutrophil engraftment, and this is kind of important and what I heard today changes some of what I was thinking in the past. Is it your feeling that if there were enough cord blood cells given that there would be no real difference to time of engraftment compared to bone marrow transplantation? This is kind of important because it may reflect on what the stem cell is like in cord blood versus bone marrow. DR. HOROWITZ: Well, I think that the honest answer to that question is we don't know because no one would ever give those few bone marrow cells to a person to be able to test it, and there are very few, if any, cord blood transplants done with cell doses that are in the range of bone marrow transplants -- DR. BROXMEYER: But you wouldn't have to go that high. DR. HOROWITZ: -- in similar patients, same age, etc. But I think that probably more cells -- that the rate of engraftment is not that different that increase in the number of cells would likely speed it up. DR. KURTZBERG: But if you look at the babies -- Donna presented data and we, in the pediatric setting, have data of babies getting doses that are between 1 and 2 X 108 cells/kg, and engraftment is faster, and that is more striking for myeloengraftment than platelet engraftment. Platelets are still delayed compared to what you see with bone marrow but not as much as they are in the adults. I know that in our series the ANC to 500 is in the 15-20-day range in that group of patients on average, and the platelet recovery is between day 40 and 50 compared to people getting a lower cell dose where it is, you know, day 60-70 for platelets. DR. HOROWITZ: The actual times to engraftment are quite similar to T-cell depleted bone marrow transplants. So, that is another way of getting at the question. DR. BROXMEYER: Yes, but I don't know if that would be a legitimate way of looking at it. I am still trying to figure out -- I mean, the closest I have seen is what Donna presented -- DR. HOROWITZ: Right. DR. BROXMEYER: -- where it was pretty quick, I mean, between 10 and 20 days, it looked like. DR. HOROWITZ: The problem of comparing, and the reason I said what I said is that the problem of comparing infants is that the infant cell doses with bone marrow transplants tend to be higher than that, and then age is an independent prognostic factor with bone marrow transplants for time to engraftment, even independent of cell dose at least in our data. So, you don't have exactly a comparable group but I think within the limits of what we know, probably increasing the number of cells will bring the times to engraftment to be very similar. The question is whether immune recovery will be the same, for better or for worse. DR. DYKELET: Claire Dykelet, CDC. I have a comment and a question for Dr. Takahashi. First of all, I would like to compliment you on looking at the different causes of outcome in your patients, looking at the different causes of death, and I am interested in the fact that you found that infection was the second most common cause of death. I think that is a really important thing to be looking at. The only way we can really start to prevent death in these patients is to figure out what the causes are and then we can start to target them individually. Can you tell me, please, what were the infections that were most common? Was pneumococcal disease a problem? DR. TAKAHASHI: I am sorry, I don't have the data in my hand, but I will let you know. DR. DYKELET: Okay, I will give you my card after. DR. KURTZBERG: I can make a comment about that. In our patients where infection is the leading cause of death it is pretty equally split between fungal, viral and bacterial. It is not pneumococcus in this early 100 days. It is gram-negatives and gut organisms for the infections, enterococcus etc., and for the viruses it is adenovirus, CMV and in our series it is flu too that has been a problem but I think that is environment -- what shall I say? -- I think that is just related to the time of year we happen to be transplanting and what is in our community. For fungal it is both resistant Candida and Aspergillus. DR. HOROWITZ: Do you think that that is different from bone marrow recipients? It is the same spectrum of infections, though perhaps a higher incidence. DR. KURTZBERG: Right. DR. WERNET: Peter Wernet, from Dusseldorf. What is your impression concerning the CMV story? The cord bloods are in all likelihood negative for CMV when donors are positive for CMV. What is the chance for increased reactivation of that latent infection of the cord blood transplantation as compared to bone marrow transplantation? DR. RUBINSTEIN: Could you rephrase -- DR. HOROWITZ: CMV reactivation in CMV positive recipients? Is that what you are saying? DR. WERNET: Yes. DR. LAUGHLIN: I will take a stab at what I think is your question. When the European group made their observation of the prognostic value of recipient CMV serologies on transplant outcomes -- and this is simply my personal conjecture or hypothesis -- I think it has to do with perhaps a hidden linkage with particularly acute graft versus host disease and then that higher incidence leading to higher incidence of chronic graft versus host disease, i.e., patients who are transplanted with positive serologies are at increased risk of reactivation and with that reactivation known interactions between reactivation of CMV and exacerbation of acute graft versus host disease. That is somewhat my interpretation. DR. WERNET: Yes, because there have been, as you said, these discrepant results without it. DR. KURTZBERG: I also think CMV is a moving target because we have managed CMV differently over the years. We used to prophylax anybody who was serologically positive. Then, I know, we went to not prophylaxing those people but following first antigen and now DNAs and only treating people when their DNA converts. I know in our whole group we have three patients who are the way Mary is describing it, the kind who have this interaction between CMV and GVH and all three of them eventually died, although one of them lived for a year and a half before she died. But I think that because we detect it differently and we manage it differently it is very hard to look at a conglomeration of data over the last six or seven years and make a cohesive story. DR. WERNET: So, would you recommend on that basis perhaps a more rigorous preventive treatment in patients receiving CMV negative cord blood units where the patients are positive, right from the beginning just because of that? DR. HOROWITZ: I don't think that there is any reason to have a difference in the policy towards CMV infection prevention in bone marrow transplant recipients versus cord blood transplant recipients, and in CMV positive recipients who have CMV negative versus CMV positive donors. DR. WERNET: Okay. DR. HOROWITZ: I think in all of those instances an aggressive policy towards early detection and preemptive treatment or prophylaxis is indicated. DR. WAGNER: It is a good question though, but we don't know the answer to it. We never did a comparative study in CMV incidence or disease between cord blood versus bone marrow. So, we don't know the answer specifically to your question but my gut feeling is that there is nothing different about the disease. Once it occurs it is very similar -- DR. WERNET: Yes -- DR. WAGNER: -- but we haven't really formally looked at that but I am writing it down! DR. KURTZBERG: I mean, the other thing to say is there is a price to pay for prophylaxis. DR. WERNET: Okay. DR. KURTZBERG: So, it is not without cost. I don't mean monetary. DR. LAUGHLIN: I would also add to that. I think that the moving target is going to move more if the experience in our program reflects that of other programs, the financial constraints, particularly in the administration of intravenous immunoglobulin as prophylaxis because of fiscal concerns. DR. RUBINSTEIN: I am not sure that this speaks to the question but yesterday Cladd showed what the situation between the donors and the recipients with respect to CMV detection is in patients -- DR. WERNET: Yes. DR. RUBINSTEIN: -- and, from that analysis it is clear that it is much more likely to be a reactivation of an existing situation. DR. COELBO: I have a question at least initially targeted to Pablo but then anyone el