U

P-R-O-C-E-E-D-I-N-G-S

11:53 a.m.

DR. RAO: So welcome to the afternoon session of the meeting. Before we can commence formally Gail will read out a statement. I'll turn it over to Gail.

MS. DAPOLITO: Good afternoon. Pursuant to the authority granted under the committee charter, the Director of FDA, Center for Biologics Evaluation and Research, appointed the following individuals as temporary voting members for the committee discussion of retroviral vector-mediated tumorigenesis in gene transfer clinical trials, Topic II for the meeting: Ms. Barbara Ballard and Drs. Rebecca Buckley, Warren Leonard, Richard Mulligan, Daniel Salomon, and Linda Wolff.

Based on the agenda, the FDA determined that there are no specific products being considered for approval at this meeting. The committee participants were screened for their financial interest to determine if any financial conflicts of interest existed. The agency reviewed the agenda and all relevant financial interest reported by the meeting participants.

The Food and Drug Administration prepared general matters waivers for participants who required a waiver under 18 USC 208. Waivers were granted to Drs. Rebecca Buckley, Warren Leonard, Richard Mulligan, And Anastasios Tsiatis.

Because general topics impact on many entities, it is not prudent to recite all potential conflicts of interest as they apply to each member. FDA acknowledges that there may be potential conflicts of interest but because of the general nature of the discussions before the committee, these potential conflicts are mitigated.

Ms. Lawton will be participating as the nonvoting industry representative acting on behalf of regulated industry for the discussions of Topic II. Ms. Lawton's appointment is not subject to 18 U.S.C 208. She is employed by Genzyme Corporation and thus has a financial interest in her employer. Genzyme

has financial associations with universities, investigators, and research foundations.

With regards to FDA's invited guest speakers for the discussions on Topic II, the following disclosures will assist the public in objectively evaluating presentations and/or comments made by the participants.

Dr. Christopher Baum is employed by Hannover Medical School, Hannover, Germany. He holds stocks in firms that could be affected by the discussion and has unrelated grants with firms that could be affected by the discussions.

Dr. Utpal Dave is employed by the National Cancer Institute, National Institutes of Health. Dr. Cynthia Dunbar is employed by the National Heart, Lung, and Blood Institute, National Institutes of Health. As part of her official Government duties she collaborates with several researchers and firms that could be affected by the discussions. She is involved in gene therapy research.

Dr. Donald Kohn is employed by the Division of Research Immunology, USC Keck School of Medicine, Los Angeles. FDA participants are aware of the need to exclude themselves from the discussions involving specific products or firms for which they have not been screened for conflicts of interest. Their exclusion will be noted for the public record.

With respect to all other meeting participants we ask in the interest of fairness that you state your name, affiliation, and address any current or previous financial involvement with any firm whose products you wish to comment upon. Waivers are available by written request under the Freedom of Information Act.

Thank you, Dr. Rao.

DR. RAO: Thank you, Gail. Since we have many new members on the panel, I'm going to ask them to introduce themselves briefly starting on my left with Dr. Leonard.

DR. LEONARD: Warren Leonard, Laboratory of Molecular Immunology, National Heart, Lung, and Blood Institute.

DR. WOLFF: Linda Wolff, National Cancer Institute, Laboratory of Cellular Oncology.

MS. BALLARD: Barbara Ballard representing SCID families along with Immune Deficiency Foundation.

DR. MULLIGAN: Richard Mulligan from Harvard Medical School.

MS. LAWTON: Alison Lawton, Genzyme Corporation and representing the industry.

DR. MURRAY: I was here before. Tom Murray from the Hastings Center.

MS. TERRY: Sharon Terry, Genetic Alliance.

DR. BLAZAR: Bruce Blazar, University of Minnesota.

DR. RAO: Mahendra Rao, National Institute of Aging.

MS. DAPOLITO: Gail Dapolito, Center for Biologics, FDA.

MR. TSIATIS: Butch Tsiatis, Department of Statistics, North Carolina State University.

DR. ALLAN: John Allan, Southwest Foundation for Biomedical Research.

DR. HIGH: Kathy High, Children's Hospital of Philadelphia.

DR. TOMFORD: Bill Tomford, Harvard Medical School.

DR. HARLAN: David Harlan, National Institute of Diabetes, Digestive, and Kidney Diseases.

DR. BUCKLEY: Rebecca Buckley, Duke University Medical Center.

DR. O'REILLY: Marina O'Reilly, NIH Office of Biotechnology Activities.

DR. WILSON: Carolyn Wilson, Division of Cellular and Gene Therapy, CBER, FDA.

DR. SIMEK: Stephanie Simek, Deputy Director, Division of Cellular and Gene Therapy, FDA.

DR. RIEVES: Dwaine Rieves, FDA CBER also.

DR. RAO: Maybe I can have the speakers introduce themselves very briefly, too.

DR. DUNBAR: Cindy Dunbar, NHLBI.

DR. KOHN: Don Kohn, University of Southern California.

DR. BAUM: Chris Baum from Hannover and Cincinnati.

DR. UTPAL: Utpal Dave, Mouse Cancer Genetics Program, National Cancer Institute.

DR. RAO: Before we start I have one more business announcement to make and that is related to open public hearing announcements. I'll just read it out verbatim.

"Both the Food and Drug Administration and the public believe in a transparent process of information gathering and decision making to ensure such transparency of the open public hearing session of the advisory committee meeting. The FDA believes that it is important to understand the context of an individual's presentation.

For this reason, the FDA encourages you, the open public hearing speaker, at the beginning of your written or oral statement to advise the committee of any financial relationship that you may have with any company or any group that is likely to be impacted by the topic of this meeting. For example, the financial information may include the company's or the group's payment of your travel, lodging, or the expenses in connection with the attendance of the meeting.

Likewise, the FDA encourages you at the beginning of your statement to advise the committee if you do not have any such financial relationships. If you choose not to address this issue of financial relationships at the beginning of your statement, it will not preclude you from speaking."

Before we open the committee session, we would like to honor Dr. Mulligan and I'll ask that he come forward.

DR. MIDTHUN: I'd like to take this opportunity to thank Dr. Richard Mulligan for his service to this committee. He actually rotated off the committee last year but has once again come to help us out as we tackle very challenging issues related to cellular and gene therapies. We really do acknowledge the time and dedication it takes to serve on this committee and we very much appreciate it. I just have a small token of appreciation for you.

DR. MULLIGAN: Thank you very much.

DR. RAO: So we have one speaker who has requested time to speak and that's Dr. Claudio Bordignon. If there is anybody else, you should raise your hand so that we can recognize you. Please make sure you stay within the allotted time.

DR. BORDIGNON: May I ask how much time do I have? Seven minutes. Thank you.

My name is Claudio Bordignon and I'm scientific data editor of a private foundation and hospital in Italy, a nonprofit foundation. I'm also associated with a gene therapy company dedicated to the treatment of cancer and AIDS. None of the data that will be presented today are sponsored by any industrial company. They are actually supported by Telethon, an Italian foundation dedicated to the treatment of genetic diseases.

The difficulty I will be addressing very briefly today but, of course, I'll be here to answer specific questions or additional clarification that relates to point No. 3 of the questions for committee discussion. Most specifically whether ADA-deficiency gene therapy treatment should be related or affected by the results that have been observed in the gamma chain deficiency trial.

I will not make specific comparison of the direct relationship. I will just like to describe to you the data, the duration of the trial, and the outcome so that the committee can make up his or her own mind on the result and the potential clinical benefit to the patient as well as the results from the basic science.

This protocol, as you may probably see from a quick look to the description chart, is very similar to the one utilized for the gamma chain. It's based on the collection and separation of CD34+ hematopoietic stem cells. The activation to favor gene transfer integration is very similar with very minor differences to what has been described before by Alain Fischer and Marina Cavazzana-Calvo and colleagues. And the transaction utilizes the additional contribution of retronectin in the number of days that are indicated here.

This is the structure of the vector. The ADA transgene is driven off by the retroviral LTR. The unique characteristic of this product is to utilize oral or IV busulfan during this period of infusion of genetical engineer cells to favor the competition between the endogenous hematopoietic system and the incoming transduced cells. After conditioning the patient is receiving the cells as described.

We have been treating with gene therapy ADA-deficient patients for the last 10 years or more. However, the trial that is described here and as described in the charts. The first one including busulfan treatment or nonmyeloablative conditioning is described here.

Here you see the characteristic of the different patients. The previous existing treatment, two or three patients were on PEG-ADA for a short or longer time. I should also thank Dr. Buckley for addressing to us one or two of these patients.

This is the age of treatment so please note there are two relatively young patients at seven and 12 months. This is the duration of the follow-up. The longest treated patient is today 43, 44 months.

This is the outcome of the treatment. You will probably right away notice that there is a difference in engraftment of genetically engineered myeloid lineage and T-cells. There is a relatively large difference in the proportion of transduced cells so the average is around 10 percent. What is most important it remains stable after the first few months.

No matter what level of hematopoietic stem cells transduced or T-cells thanks to what we interpret as positive selection due to the presence of toxic substrate that are detoxified and metabolized in the genetically engineered cells all T-cells go to 100 percent transduction.

This is the level of immune reconstitution. Here there are several characteristics. Since I don't have the time to go through the slides, I will summarize here the points. What is relevant is that in essence all the patients received -- obtained or achieved immune reconstitution. That is full in all the patients with the exception of one. The patient, you may have noticed, received a relatively low dose of genetically engineered cells.

What is resulting as the control of the normal expression of the transgene is restoration of thymic function, normal in vitro T-cell responses to mitogens and antigens, diversified T-cell receptor repertoire, normal development of B and NK cells, improvement of serum immunoglobulin, system production of specific antibodies following vaccination.

The clinical benefit from the patient is this three-year-old patient is at home living a normal life. In different parts of the world prophylactic drugs have been discontinued in all patients. No severe infection. Normal neurological development, normal hearing. These are some site complication of ADA-deficient patient. No adverse effect or toxicity has been observed. None of the patients required the use of PEG-ADA after treatment.

We have also analyzed -- this will be my final slide. We have also analyzed a number of integration and gene expressions specifically in relation to the adverse event observed in France. The first 370 integrants who have been observed can be summarized here for their characteristics. They are highly heterogeneous, polyclonal integration of circulating T-cells, fewer integration in granulocytes.

Frequency of integration inside the gene has been reported that 36 percent was observed in the study. Genomic regions close to transcription start sites are preferential sites of integration. There has been no clonal expansion observed throughout the time. No preferential gene heat retroviral vector, no significant difference in the pattern of gene expression in T-cells compared to active control.

I would like to conclude here that clearly in our experience for this trial gene therapy has provided the best results we have available as compared to PEG-ADA and not haploidentical transplantation. Since Don Kohn will be speaking after me and he will come back to this issue. I will leave up to him the details of the discussion.

Thank you very much for you attention.

DR. RAO: Thank you. If there are no specific questions, I'll ask Carolyn to go ahead.

DR. WILSON: Thank you. I want to welcome the members of the committee as well as our guests of the committee for returning to this discussion of retroviral vector-mediated insertional tumorigenesis. This is, in fact, the third time we've discussed this topic with this committee. The previous two times were in October of '02, February of '03.

Since that time because there are also new members of the committee, I want to actually start my update with a brief review of a mechanism of retroviral insertional tumorigenesis, review what was previously known about the X-SCID gene therapy clinical trial in France, and then initially provide you just a brief update of what's new, review how we've responded both previously and more recently.

Then, finally, come back to the February '03 meeting of this committee in order to discuss the recommendations that this committee made to us in the context of actions that have been subsequently taken.

Then I am going to return to the gene therapy clinical trial in France with a more detailed update and then finish with an introduction of the other presentations for today and the questions for the committee that we will be returning to towards the end.

I want to start by thanking Professor Jean-Hughes Trouvin and Alain Fischer and Marina Cavazzana-Calvo who cooperated with the French Regulatory Authority AFSSAPS. They've provided to me up-to-date details regarding the third case of leukemia in their clinical trial. I wouldn't be able to present this information to you today without their cooperation and help.

It's long been known that gamma retrovirus infection can, in fact, cause activation of endogenous genes. The primary mechanism for this, and this is shown in the proviral form of a retroviral vector. Shown at either end are repeat elements called LTRs, or long-terminal repeats, which contain at the 5-prime end an element known as the U3. This has a very strong enhancer and promoter.

It is known that in the context of when this provirus is integrated into a chromosome that this U3 element may cause gene activation at quite far distances for wild-type murine retroviruses as far as 300 kilobases away. But it can also mediate read-through transcription at the three-prime end where downstream genes may be activated.

In addition, just the mere integration event itself may also cause disruption of a genetic locus. Any of these types of events may lead to dysregulated gene expression. This in and of itself may or may not have any phenotypic effect on the cell or any obvious effect in terms of a clinical change.

However, in some cases it may in combination with probably a multi-step process eventually lead to tumorigenesis and that's really the topic we are here to discuss today.

So to come back to where we were two years ago when we came to this committee, at that time the clinical trial in France that had treated 11 children using autologous CD34+ cells that had been modified ex vivo using a retroviral vector encoding gamma-c. I'll also be referring to it with the shorthand on my slides. This is the gene that is defective in x-linked severe combined immunodeficiency.

At that time 10 out of 11 had shown both clinical and laboratory evidence of both engraftment and immune reconstitution as well as health benefits. Unfortunately, two out of 10 children, as we now know, also develop leukemia. The results of the molecular characterization of those leukemias were both presented to this committee two years ago but have also subsequently been published in Science in 2003.

I won't go through them in great detail but just wanted to briefly remind the committee that both patients, patient four and patient five, showed retroviral vector sequences integrated within a locus called LMO-2. Patient four was in reverse orientation within the first entron while patient five was in the 5-prime untranslated region.

In addition, the investigators demonstrated that unlike normal T-cells shown here, which are negative for expression of LMO-2 RNA, both patient four and patient five leukemic cells were transcriptionally activated for this locus.

So basically to summarize then, what was shown is that the retroviral vector had integrated into a genetic locus called LMO-2 and this resulted in transcriptional activation of this locus. In fact, LMO-2 is also associated with a subset of childhood T-cell lymphoblastic leukemia.

Together these evidence are very suggestive that, in fact, the leukemias were a direct result of the retroviral vector-based gene therapy. That was the conclusion of this committee at that time as well as a number of other advisory committees who have discussed this topic subsequently.

Now, in January of 2003 we sent letters to retroviral vector sponsors. I'm just summarizing what was the content of those letters. We sent three different categories. We sent one to sponsors of trials that were using retroviral vector-based therapy for treatment of skin, as well as any other clinical indication that used active hematopoietic stem cells.

This letter was a clinical hold requesting sponsors to revise their informed consent as well as to develop methods to monitor for clonality of vector integration.

Those that also used hematopoietic stem cells as a target but were inactive at that time were sent a similar letter just notifying if they were ever interested in resuming a clinical trial under that IND that they needed to also address these same two issues. Then other retroviral vector clinical trials that use different target cells were sent a letter where we just recommended that they also follow these same two steps.

We also asked all sponsors to provide a very important risk benefit analysis in the context of their clinical indication, alternative therapies, and so on. Then we evaluated each response. In a few minutes I'll come back to those responses.

Now I wanted to just briefly also tell you what has happened in the subsequent two years in this clinical trial in France. In May of 2004 the French Regulatory Authority allowed the clinical trial to resume with the following changes to their protocol. They had a maximum cell dose of 10 million gamma-c positive cells per kilogram.

The children who participated in this protocol had to have no family history of childhood cancer, no cytogenetic abnormalities, and had to have at least one prior infectious episode, as well as having a minimum age of six months. Since the resumption of this protocol, only a single patient has been treated.

As many of you may have heard, patient four, the first child who developed leukemia, has since had a relapse and died in October of last year. Patient five, his status is that he's still in complete remission since completing chemotherapy.

Now, unfortunately, this is the subject of today's discussion, a third patient, patient 10, has now presented with a T-cell lymphoproliferation.

So in response to these most recent events, in the end of January the FDA responded by placing three INDs on clinical hold, two that are using retroviral vectors to treat X-SCID, one for the treatment of ADA-SCID. We asked these sponsors to revise their informed consent document indicating that an additional child had developed a severe adverse event as well as to notify their IRBs.

In addition, we sent a letter to all sponsors that use retroviral vectors to make sure that they were aware of these new events, as well as a letter to all IRBs that are responsible for regulating these types of clinical protocols.

I want to also give a blanket thanks to the three sponsors of these INDs. In the case of X-SCID Dr. Ken Weinberg of Children's Hospital, LA and Dr. Harry Malech of NIH-NIAID for. For ADA-SCID Dr. Don Kohn of Children's Hospital, LA. These three sponsors have been very cooperative and willing to allow us to present data from their IND in this public forum and that is a really very helpful thing for the field.

Now I'm going to turn to the discussion two years ago, what recommendations you made as a committee to us and our actions. I'm using this color coding hopefully so people won't get too confused as I go through this.

At that time the committee had a unanimous vote in favor of allowing the use of gene therapy in children only when there were no alternative therapies available when the clinical indication was either X-linked SCID or IL-7 and JAK-3 deficiencies.

As I mentioned, there were two US INDs for X-SCID at that time who had been placed on hold. They subsequently responded with the following protocol changes that allowed them in the interim to resume treating patients. I wanted to briefly review their changes so you are aware of this.

Dr. Malech is treating only patients that had previous bone marrow transplants but still show persistent T and B lymphocyte impairments, as well as a compromised quality of life. He's only treating older children that are at least two years old. Initially he proposed doing this on a case-by-case review of each patient by both the IRB and the FDA and has subsequently been told by both institutions that is not necessary.

Dr. Weinberg is only treating patients who have no HLA-matched sibling and while they may have a haploidentical transplant donor, they have preexisting infectious complications, for example, or other conditions that may make them at high risk for receiving such a transplant. They have also placed a cap on their cell dose of 20 million cells per kilogram. With their average transduction efficiency this is approximately 5 million gamma-c+ cells.

So for ADA-SCID the committee actually expressed a wide spectrum of viewpoints so I just wanted to use the following slide to give you a flavor of some of the major points that were raised by committee members. Some felt there was no biological difference between the ADA and X-SCID suggesting that the risk would be the same.

Related to that is the conclusion by some members that the transgene for X-SCID gamma-c doesn't have a role in the leukemogenic process. Then it was also pointed out that because the cell dose is typically quite low compared to X-SCID when you do ADA-SCID, bone marrow CD34 transduction, that this would reduce the risk accordingly.

But then others thought the transgene in X-SCID actually does have the potential to be oncogenic and that the ADA-SCID transgene is simply metabolic and, therefore, we are really looking at apples and oranges and that there may be a greater potential in X-SCID that the gamma-c may be providing a second hit or somehow potentiating the leukemogenic process.

So obviously then no consensus really came out of that discussion. Some members thought that the clinical hold should be removed. Some thought that it should only -- gene therapy should only proceed in children who fail PEG-ADA or fail bone marrow transplant. Essentially those have no alternative therapy. But then it was also pointed out that if you don't have the selective advantage of not being on PEG-ADA with gene therapy, then you lose the potential for selection and efficacy.

As I mentioned, Dr. Kohn had an IND for ADA-SCID and he made the following changes to respond to these concerns. He is only treating children at least six months old. He will only treat children who either come off of PEG-ADA two weeks prior to treatment or have never received PEG-ADA and then maintain that condition of no PEG-ADA following the transplant.

Then the third point is more related to really efficacy than a safety issue which is to institute use of a conditioning regimen busulfan treatment five days prior to transplant.

then, as I mentioned, there were clinical hold letters sent to all other protocols that used ex vivo retroviral vector transduction of matopoietic stem cells.

The committee thought that this category still had a significant risk similar to X-SCID as long as there were adequate numbers of transduced cells infused. They thought this was an important consideration that should be looked at in the context of risk benefit analysis and asked sponsors to analyze the probability for their trial.

But, then again, other members thought their risk would not be the same as it is with X-SCID. Again, this comes back to the diversity of viewpoints about the role of the gamma-c gene in the leukemogenic process. So finally, then, some thought that there would be -- it would only be suitable when there is no alternative effective therapy or standard therapy had failed.

So the consensus vote was that 18 out of 19 members asked that we reevaluate all the trials on hold and look carefully at the risk benefit of using gene therapy versus existing alternative therapies for that particular clinical indication, as well as making sure there was appropriate informed consent changes.

The one member who voted against this encouraged FDA to request an analysis through sponsors of the probability of a similar adverse event occurring in their individual trial. So we have implemented this recommendation by evaluating each sponsor's response to the January letter looking at the informed consent revisions, looking at their response of how they are doing clonality analysis and, of course, evaluating carefully the risk benefit issues.

I can't go through each and every IND that was affected by that clinical hold but what I can do is sort of summarize for you an aggregate, the current administrative and IND status for each of these INDs. After that January '03 letter four sponsors withdrew their IND. These were older INDs that have finished treating patients.

Another four actually responded to the letter and it was an adequate response, went back into effect but then subsequently inactivated along with another IND that inactivated. then we still have 19 active INDs.

Of those nine have responded to the clinical hold letter adequately and are in effect. Then another 10 are still on hold and I want to break those down for you. Three of those 10 actually are the three SCID INDs that had come off of hold but were then placed back on hold in January of this year.

Six out of 10 have never responded to the clinical hold letter. Again, these are really older INDs that have finished treating patients. Then one sponsor responded to the letter but it was inadequate and they are still on hold. I also just wanted to mention that since January of '03 we've had 15 new INDs submitted using retroviral vector and the breakdown of those INDs is shown here.

I want to finish this portion of my talk by just briefly explaining some of the other recommendations that the committee made not related necessarily to the INDs that were on hold. One area that the committee recommended was to look at how to improve vector safety and they identified the following issues as being important, suicide vectors, insulator elements, looking at the issue of cell dose and number of integration sites.

I wanted to mention a collaborative study that the FDA is in the process of initiating through the National Toxicology Program. This is going to include investigators from FDA, NIH, and academia. Academic investigators are led by Chris Baum in Cincinnati and David Emery at the University of Washington.

What we propose doing through this program that has the resources to do very large-scale animal studies is to develop a systematic approach to look at various vector modifications or transduction modifications that may affect the risk of vector-mediated tumorigenesis.

What we want to do is do this in a study of sufficient size to give a 90 percent confidence interval and a negative result. We think this would be an important study to do (a) to gain information on whether any of these changes will have an effect on the risk. It also allows us to get more data about one mouse model and to see whether or not this may provide a good mechanism to screen new vector modifications for increased safety.

Also, in conjunction with that, the American Society of Gene Therapy, BIO and PhRMA co-sponsored last year a workshop and long-term follow-up of participants in human gene transfer research. This workshop discussed the scientific preclinical, clinical, social, legal, and ethical issues related to the long-term risk of gene therapy. Obviously the issues associated with retroviral vectors and the delayed risk of tumorigenesis were important to that discussion.

Two other areas I wanted to mention. The committee cautioned us that we shouldn't extrapolate one adverse event from one trial to all others just because they may use the same vector in the same target cell. So with that in mind, I did want to mention that we only placed SCID INDs on hold this time after this third child became known to us, not any INDs that aren't in SCID.

Then, finally, the committee recommended that we develop a mechanism to study the number of integrants that would provide reduced risk. I wanted to just briefly mention a committee within FDA that I'm a part of along with Dan Tekefman in DCGT, Steven Anderson who is Associate Director for Risk Assessment in the Office of Biostatistics and Epidemiology.

I want to present to you just one aspect of this model that we were working on which is to use a computer model to estimate the number of vector integrations and LMO-2 just as a way to get started. What Steve Anderson did is he used both the coding region of LMO-2 as well as 10.0 kb upstream and downstream.

And then using actual data from the study of Shawn Burgess published by Wu et al. in Science in 2003 looked at actually 1.0 kb increments and estimated the probability of insertion in each of these regions. Then used this model to look both by a random insertion probability and then what he calls a gene-biased insertion.

From that if he looks at the number of vector integrations in LMO-2 per million cells based on looking at different numbers of vector copies per cell, you can see -- oh, sorry. This got a little messed up. Even at a single vector copy per cell with the random model we can find 20 cells within that million that would have a vector integrant in LMO-2. Obviously the numbers increase with increasing vector copies and with the gene biased model they are even higher.

Then we asked what would be the probability then in the treatment of 1 million cells, one or more cells with that vector integrant within LMO-2 would actually engraft. For this we used an assumption based on data from Cindy Dunbar's nonhuman primate studies that only 1 in 100,000 cells would engraft. Now we're asking what is the likelihood, for example, that these 20 cells in the million will end up being that 1 in 100,000.

As you can see, these numbers are quite low so that this is saying what is the probability that if a patient were to receive a treatment of 1 million cells that one or more of those cells would have a vector integrant in LMO-2. So this tells us that 1 in 5,000.

In other words, you would have to treat 5,000 patients before you would see a single vector integrant in LMO-2. This clearly is not consistent with the clinical data from the French trial clearly indicating that this process of engraftment is not a selection neutral process, that there must be some advantage for having an integrant in LMO-2.

So now I want to turn to a more detailed update about patient 10 from the clinical trial in France. Again, I really need to thank Jean-Hughes Trouvin at AFSSAP for very kindly providing this information to us.

This patient was treated at eight months of age. He received 11 million CD34 gamma-c+ cells per kilogram. The time point previous to the lymphoproliferation which was only two months earlier. There was no evidence for any clonal expansion. But at approximately 33 months the patient presented with cervical lymphadenopathy and rhinopharyngitis.

Upon subsequent investigations they found that he also had an enlarged mediastinum and that this was, in fact, associated with T-cell lymphoproliferation. Blast cells were found in the blood, lymphnodes and bone marrow. These cells that were proliferating are both CD4 and CD8+ indicative of immature thymic cortical thymocytes. They are gamma-c+ indicating that they are vector positive.

While the colonality and integration site are under investigation, so far the data suggest it isn't due to an integration in LMO-2. Earlier timepoints that were analyzed have no LMO-2 integrants and the lymphoproliferative cells don't have over-expression of the LMO-2 protein. Fortunately at this time the patient has responded very well to steroid treatment and is in complete remission.

Now, I wanted to just compare this third patient's profile with the previous two. Obviously I can't include every detail of these clinical events but I wanted to just focus on a few issues. Especially the age and the dose because when we discussed these two patients two years ago, two things that seemed to be unique among patients four and five compared to the other patients that were treated were their young age of one and three months and their relatively high cell dose.

Patient 10 is somewhat older. He's a nine-month-old child and he received a lower cell dose than patients four and five, although I will say this is still higher than the median dose in this clinical trial which was 4 million cells.

The time of the T-cell proliferation is eerily similar. All of them developed this effect at approximately 30 to 34 months post-transplant. The phenotype of the T-cells is somewhat different in each case. The first were gamma-delta+ T-cells with a single T-cell receptor. The second had alpha-beta+ T-cells and actually had three T-cell receptors within those leukemic cells indicating that the proliferative event occurred prior to T-cell receptor rearrangement. This third child has a clonal T-cell receptor Vbeta5.

As I already mentioned, the first two patients had the integration in LMO-2. The third has actually three sites of vector integrants. This has been confirmed by both LAM-PCR as well as TaqMan PCR but the exact sites of integration has not yet been determined.

So to just counterbalance this information, I also wanted to mention that there are still nine other patients in this protocol, six of whom have had no intervention other than in gene therapy and they are still healthy and fine including patients one and two who are now out almost six years. Again, our sustained -- having a sustained clinical benefit with no treatment other than gene therapy.

Patient three who is noted here as having failed gene therapy is actually somewhat interesting. He received BCG vaccination which resulted in splenomegaly. Although they didn't find any of the gamma-c+ cells in his peripheral blood, when he had a splenomectomy it turned out that all of the gamma-c cells had been aggregated in his spleen. Subsequent to that he had a bone marrow transplant and he is healthy.

Patient nine had a very low cell dose and apparently T-cells were always low and then as they started to decrease even further, he ended up having a bone marrow transplant. Then the third patient the gene therapy hasn't worked in was a much older patient diagnosed at the age of 13. He was treated at the age of 15.5 years and even in spite of a very large dose he has not shown any evidence of T-cell reconstitution. The investigators in France think this is due to poor thymic function.

I wanted to mention that the French regulatory authorities in their report to me said that the regulatory status of the X-SCID clinical trial in France is that they are on hold. This is at the request of the investigators. In their report they note that the risk of insertional mutagenesis is clearly identified today with the combination type of vector and the nature of the transgene.

This risk warrants a necessity of reconsideration on the protocol and method used for this SCID-X1 gene transfer. They also mentioned that the use of safer design vectors is to be envisaged in an attempt to reducing the insertional mutagenesis risk.

I'm not exactly clear what they mean by this last statement in terms of whether or not this is something that they would require before the protocol would proceed but I just wanted to mention that this is something that is in their mind.

Now, one question that has come up at this committee discussion both times is what is the likelihood of leukemia in SCID patients who receive bone marrow transplant. To try to address that question, Dr. Rebecca Buckley, who is also here today as a guest of the committee, can speak to this better than I can if there are additional questions later. She published last year in the Annual review of Immunology a very nice survey of 132 SCID patients who had received either haploidentical or HLA-identical transplants.

Of those 102 have survived and 68 out more than five years. I think what is notable here is while 30 of these children have subsequently died, the majority are from infectious complications. Then there are a few from some odd clinical syndromes. None of them have died from hematologic milan menses.

Then another issue that was also discussed is the role of the gamma-c transgene product. I wanted to just mention the published data from Dr. Fischer's group Hacein-Bey-Abina in Science of 2003 where they showed that gamma-c was not overexpressed in the leukemic cells.

They sequenced the entire vector in the gamma-c transgene as well as the vector were all as expected. Then another marker for potential disregulation by gamma-c which is constitutive activation of the JAK-3 kinase was absence in their leukemic cells.

In addition, as I'll mention in a minute, there are other clinical trials in X-SCID and at this time there have been no observed hematologic malignancies. This perhaps argues against it but there could be another reason for this.

Then more recent data that was published last year from Utpal Dave who will talk later today is the identification of leukemias in a murine retrovirus model that maps oncogenes that where some leukemias as developed that have insertions in both gamma-c and LMO-2 suggesting a potentiating role for these two gene products.

As I mentioned, there are other gene therapy clinical trials in X-SCID. Adrian Thrasher in the UK has been doing a protocol. He published the results of this last year in the Lancet. Dr. Ken Weinberg at Children's Hospital is doing a protocol but at this time no children have been treated. Dr. Malech has treated two children. The longest time to follow-up is in Dr. Thrasher's protocol and his is out just a little over two years at this point. It's hard to know whether this is an issue.

We looked very carefully at specific details of the transduction method and culture conditions of the French clinical trial versus these other two protocols to see if there might be something unique to this protocol that could explain the difference. There are some specific media additives that are present in the French protocol that aren't in the other protocols.

They also use an amphotropic murine leukemia virus envelope as opposed to the other two which use gibbon ape leukemia virus. This might have some impact on the target cell that gets transduced but it's really unclear. Then again, as I mentioned, Dr. Fischer's protocol has the longest time to follow-up with patients out actually beyond five years. What is hard to know right now without the other protocols having longer times to follow-up is what is truly the significance of any of these characteristics.

Finally, I just want to mention that we'll have four guest speakers today. The next three speakers will be presenting data to you that is relevant to this issue from animal models. Dr. Cindy Dunbar will present some information from her nonhuman primate studies of retroviral vector transduced hematopoietic stem cells.

As I mentioned, Dr. Utpal Dave will be talking about his studies with mouse retrovirus induced leukemias. Then Dr. Chris Baum will present data from his mouse model of murine retroviral vector transduced hematopoietic stem cells.

Finally, Dr. Don Kohn will present an update from his clinical trial experience using gene therapy to treat ADA-SCID.

Finally, I want to just very briefly paraphrase the questions so you have these in your mind as we move forward through the afternoon. We will obviously come back to them in more detail towards the end.

The first question is asking, "What incidence of leukemia in clinical trials using retroviral vector mediated gene therapy for the treatment of X-SCID meets the following? Human subjects are or would be exposed to an unreasonable and significant risk of illness or injury." We are really asking the committee to give us some idea of whether or when there might be a threshold beyond which gene therapy in its current form may not be a useful method to treat X-SCID.

If or when we get to that point, we would also like the committee to discuss how investigators could reduce the risk to subjects. For example, changes in the dosing paradigm or vector design. We also want the committee to discuss risk benefit considerations of gene therapy versus alternative treatments in ADA-SCID related to X-SCID. This is in some aspects coming back to the question of whether or not there is something specific to X-SCID in the transgene product, as well as other clinical indications.

Then the last question is a little bit more forward-looking and may seem somewhat off topic but is related because we know virus vectors can transduce at much higher efficiencies than current gamma retroviruses that are mostly used. We really recognize that we will be faced with products which will have much higher vector copy numbers per cell. We would like to have some feedback from the committee as to whether or not a limit should be considered and what kind of limit.

With that, I thank you for your attention. I would be happy to answer any questions for clarification at this time or we can move on to the next talk.

DR. RAO: I would suggest to the committee that unless there is some required clarification that we move on to the speakers and listen to all of them before we ask any questions or go into discussion. Our next speaker is Dr. Dunbar.

DR. DUNBAR: Ready to start, Gail? Okay. I guess I'm supposed to talk about conflicts. I don't believe I have any except I one time did a consulting for Dr. Bordignon and co-workers, company named MolMed. I do own stock which currently is felt to be a conflict of interest period. I would like to thank you for inviting me here to talk about our experience in the nonhuman primate model.

I actually didn't put any background slides in and I realize perhaps that was a mistake but I'm just going to briefly state that the rhesus macaque nonhuman primate model has been a model we've used for approximately 15 years in the hematology branch initially with Art Nienhuis and then subsequently with my group.

It has been, we believe, very useful in terms of its predictive value studying hematopoietic and studying gene transfer because its life span hematopoietic demand calculations of hematopoietic stem cell numbers and output as well as the ability to use reagents that are identical to human clinical trials such as human cytokines and human monoclonal antibodies we believe has made it a very relevant model for gene therapy.

In terms of efficiency of gene transfer, the disappointing results we had in the early to mid-1990s in human trials were mirrored using the same conditions and same vectors in the rhesus clinical trials. Over the past six years we've gotten much higher levels of gene transfer efficiency with different transduction conditions and just tweaking some aspects of the ex vivo transduction conditions and we've gotten much better gene transfer efficiencies as you'll see. That's good but it also may lead to some concerning side effects.

So I'm going to present primarily the data on one animal today that developed a tumor recently. We also have a large database on retroviral insertion sites and lentiviral insertion sites in the rhesus macaque long-term repopulating cells that I won't have time to talk about today but was published in PLoS with the first author Hematti approximately two months ago so I'll refer you to that for our overall database.

Monkey 96E113 was born in 1996 and is a sort of adolescent monkey in 1999, underwent a gene transfer protocol. This was done by Art Nienhuis, Pat Kelly, and Elio Vanin in collaboration with our monkey unit. It got 1,000 rads of TBI. The target cells were G and SCF-mobilized PB stem cells.

The CD34+ cells were enriched and then transduced with a standard retroviral vector and MCSV backbone as you see here. It included for transgenes a CFP IRES, internal ribosomal entry site, and then a modified DHFR drug resistance gene.

It was pseudotyped with RD114. It was not ampho or GaLV. It was a feline virus pseudotype that Dr. Nienhuis and coworkers were very interested in at that time. It was reinfused with 20 million cells per kilo and at the time of reinfusion 55 percent of the 34 positive cells expressed GFP.

The animal did have prompt engraftment. I will point out that this animal, the producer clone that was utilized to transduce these cells was quite unusual in that in prior NOD/SCID studies despite very high gene transfer in a good number of 34 positive cells, there was decreased engraftment of human cells that were transduced in the presence of this producer clone vector in NOD/SCID mice despite, again, increased gene transfer efficiency.

It appeared actually that phenotypically the cells were induced to differentiate by exposure not to the vector but by something that was being secreted by this particular producer clone of 3T3 cells that were producing vector. I say that because of the somewhat unusual characteristics of the engraftment pattern in this animal.

This was the marking level in the animal in the proliferal blood. This has been published -- actually, I'm sorry. It's not blood, it's PBMC by Pat Kelly, et al. in 2000. The animal had extraordinarily high and really quite persistent levels of GFE positive cells especially in the myeloid elements of polyneutrophils and monocytes here up to 70 or 80 percent that was stable out to about 6 months and then began to fall.

The level of marking and red cells and platelets was somewhat lower but the expression of GFP in those lineages does not necessarily reflect the gene transfer efficiency in those lineages. Lymphocytes remained somewhat lower throughout. However, by about one year after transplant the animal had decreased again to levels of about two to three percent.

Now, this animal when a southern blot was done on peripheral DNA in the blood and bone marrow at the time of the first year after transplant, it was interesting because even cutting with an internal cutter that cuts once within the vector, there appeared to be basically two clonal bands that accounted for the vast majority of the marking.

Normally on southern blot if you cut with an internal cutter and it's polyclonal reconstitution you see a smear or nothing. In this animal we saw these two bands, both in the blood and the marrow. Actually the presence of the bands correlated well with the high-level marking that had been seen here.

Over time as we tried to work up what was going on with this seemingly almost clonal reconstitution, my laboratory performed inverse PCR which was sort of a precursor technique to LAM-PCR to identify insertion sites. It's a very inefficient technique and it only works very well on individual colony-forming units that were grown out of the marrow one-year post transplant.

We identify two dominant clones and sequences appear to be present in chromosome 15 and chromosome 9. Looking at the restriction sites, at least in the human genome which we found very homologous and quite useful in terms of mapping the money insertions, the band size predicted to result from these two integrations was what we saw on Southern so we don't know for sure that these were the same two integrants but the evidence was very strong in terms of the predicted band size.

Dr. Vanin designed a allele-specific primers with one primer in the provirus and one primer in the genomic DNA outside in chromosome 15 and chromosome 9 and we did a allele-specific PCR and found again the presence of these clones to match what we had seen here in Southern and also here on the percent cells in the circulation.

The vast majority, I'll say again, of marking from the first six months was attributed to these clones and they both track together by PCR and Southern so it was felt that they were two inserts and one progenitor cell. When individual CFU were screened by the allele-specific primers the majority were positive for both integrant 1 and integrant 2. But a minority of CFU were found that had only integrant 1 which will become relevant when you hear later what we found in the tumor.

This animal actually because it had received the vector that contained the drug resistance gene was treated for one cycle with trimetrexate and NBMPR, basically the agent that is designed to inhibit endogenous DHFR. It got one cycle for five days in 2001 so that was two years after transplant and about 18 months after the very high level marking had disappeared.

This animal had a very, very transient enhancement of the percentage of granulocytes shown here in black and monocytes shown here with the cross-hatching after treatment with the trimetrexate. It really had gone back to baseline by about 16 days to 21 days after the treatment. This work was published by Persons et al. in Blood last year. So this may be irrelevant so the animal did get one cycle of a cytotoxic drug.

However, trimetrexate and methotrexate are actually not thought to be particularly strong mutagens in terms of causing secondary AML, secondary myelodysplastic syndrome, nowhere near the kind of risk of secondary leukemias, for instance, that you see with alkylating agents. This was the only cytotoxic drug it got.

The timeline for the next three years, the animal was basically on our long-term follow-up study. It was being checked in terms of flow cytometry for GFP and for blood counts every six months. Basically its blood counts were stable and normal over this time. The percent GFP+ cells in the blood was also stable.

Most recent before the event I'm going to talk about was June last year. Granulocytes were still one to two percent, lymphs still two to four percent. It had a physical exam every six months and the most recent full physical was in July of '04. It's weight was stable at that point and the exam was normal.

The adverse event occurred in September of last year. In early September the animal was noted to be lethargic and anorexic. X-rays were consistent with a partial bowel obstruction and it was felt by the vet at that time to be consistent with an intestinal intusseption torsion/telescoping that these animals can get with some regularity if they eat too much Monkey Chow.

At that point the white count was 14,000 and the hemoglobin was normal. It was taken to the OR because conservative treatment was not effective in terms of relieving what appeared to be severe abdominal symptoms. At exploratory laparotomy the vets went in just thinking they were going to do a simple procedure and instead found a left renal mass that was causing intestinal blockage and was up against the tail of the pancreas.

At that point this was going on in an off-campus site and, unfortunately, I was out of the country and the left kidney was removed. The was an attempt made to salvage the bowel and pancreas. Tumor samples were frozen as well as being placed in formalin but they didn't actually take tissue to try to disrupt it and sort, for instance, four individual cells and there was no cytogenetics done on the tumor itself.

Basically this is the kidney that was removed at the time. This pole here is the one that was up against the bowel and the pancreas and had adhesions. The morphology here is more normal but, as you'll see histologically, basically the entire kidney was infiltrated with tumor cells that was somewhat hemorrhagic as well.

Unfortunately, despite the best efforts of the animal care facility to keep the animal alive, the animal developed evidence for sepsis, pancreatitis, perhaps not surprisingly given where the surgical site was, and catastrophic bleeding at the operative site. It had a white count of 43,000 with a left shift -- okay. So immature cells like bands and myelocytes in the periphery but no actual blasts. No leukemic looking blasts. Hematocrit was 19 consistent with bleeding. Platelets were normal.

On chemistries, the animal had developed renal failure and liver inflammation. We took blood for molecular studies and flow cytometry. The percent positive GFP cells in the blood had increased. The overall percentage was about 30 percent. If you did a small lymphoid gate it was 3.7 percent. Clear granulocytes was 16.7.

They didn't do at that point surface phenotyping to look at a monocyte population, for instance, or some of the immature myeloid cells. I think that gate would explain the overall higher percentage of about 30 percent. The blood smear on that day again showed a left shift, monocytosis, but no clear circulating leukemic blasts.

Unfortunately, the next day the animal was found dead in its cage. It's died overnight, at least several hours previously so there was quite a bit of autolysis on the organs at path.

On autopsy there were tumor infiltrates in the kidneys, liver, pancreas, spleen, nodes, and the choroid plexus of the brain. I'll show you those in a minute. Evidence of pancreatitis and sepsis. There was left-shift in the blood and the bone marrow but there were no clear collections of the tumor cells of the immature blasts that they found in the other organs.

History histology is shown here. Here you can see the normal kidney tubules scattered about but between the kidney tubules there was a monomorphic very immature looking infiltrate of large cells here shown on higher power.

Our very helpful hematopathologist, Stefania Pittaluga and Elaine Jaffe saw these slides and thought it was a lymphoma, large-cell lymphoma. But basically doing special stains the lymphoid markers were all negative and the markers that were positive were those from myeloid cells.

This is myeloperoxidase and you see here the kidney tubules themselves were negative from myeloperoxidase as you would expect but the tumor cells were positive. We also stained with MIB1 which is a proliferation marker and, as you can see, it was very strongly positive so this was a fast-growing tumor.

So other lineage-specific stains, CD45, pan-leukocyte antigen was positive, myeloperoxidase positive, CD68 macrophage-like marker was positive, CD3 T-cell negative, CD20 B-cell negative, kappa/lambda negative so myelomonocytic tumor infiltrates.

We found these infiltrates in the liver, subcapsular like this as well as surrounding some of the portal triads. Again, they are myeloperoxidase positive. Interestingly in the brain in the white matter there weren't infiltrates but there was marked infiltration if the choroid plexus. Here you also see positive with myeloperoxidase staining.

So this would fit the definition of what we would call a myeloid sarcoma originally called chloroma. Basically this is defined by extramedullary meaning outside the bone marrow, outside the blood tumor mass or masses that are nonleukemic by definition. If you find these cells in the blood, then it's leukemia. If you find them not in the blood, it's called a chloroma or myeloid sarcoma.

What actually defines why some cells circulate in the blood and others seem to stick in organs is not entirely clear. This is relatively rare within the human spectrum of myeloid tumors, probably less than two percent.

The majority of patients that present with a myeloid sarcoma, say in one site or multi-focal, they develop acute leukemia within one to two years so it seems to be, perhaps, an immature cell that goes somewhere and transforms in situ and is able to proliferate sort of outside the normal microenvironment and eventually takes off all over the body.

Now, histologically it's often confused, for instance, with aggressive lymphomas or small round cell sarcomas until you do special stains but with special stains it's easy to say it's not a lymphoma obviously.

Okay. Now we get into the relationship with the retroviral vector. There's a lot of data I'm going to be presenting. Some of it is really a work in progress so I would hope that you wait for the final story until we've completed all the molecular studies.

It's important to point out there are no prior reports of this type of tumor in rhesus macaques in all that I've been able to find. The incidence of myeloid leukemia in rhesus macaques is extremely low. There are very, very few reports. On the other hand, I'm not sure how many rhesus macaques have been allowed to live for long lifespans. At least in humans myeloid leukemias are more common in the elderly than they are in younger patients.

It's interesting that there was a sudden increase in circulating GFP+ cells in the blood at the same time but without leukemia and we'll come back to this again at the end in terms of my hypothesis about what's going on.

Now, obviously the first thing to look at is was the vector present in the tumor. Quickly looking at immunohistochemistry for GFP it's been difficult. Our hematopathologist with out best staining believes that the tumor cells are very weakly positive. There are circulating blood cells here that are quite strongly positive but the tumor cells themselves appear to be weakly positive.

You can look at tubules next to the tumor cells. There does seem to be positivity but it's not hugely convincing in terms of the level of expression but we know that transients can certainly be turned off depending on where the integration sites are in different tissues.

Molecular analysis of the tumor. We looked at the bulk tumor but because of the infiltrating nature of the tumor cells in the kidney and also the fact that the blood was positive for vector we felt it very important to enrich as much as possible specifically for tumor cells from the kidney for the molecular analysis and then compare those enriched tumor cells to the adjacent kidney that didn't have a tumor and hopefully control that way for the amount of blood infiltration.

We did scratches. This is kind of a way to get more DNA than laser capture but it allows you to enrich for tumor cells to greater than 95 percent when you have these very patchy infiltrates that we had. We did these on the kidney. We also did these on the liver. As you remember, the liver had very intense infiltration right under the capsule and it was quite easy to do the scratches and get greater than 95 percent enrichment for tumor cells for the DNA.

So we did copy number assessment via universal vector TaqMan PCR helped by Uimook Choi in Harry Malech's lab. The copy number for the enriched tumor cell scratches from the kidney ranged from 1.4 to 4. Again, these were very small samples. We ran them three different times and got levels in that neighborhood. But the copy number was also greater than one in blood.

The copy number increased in enriched tumor DNA compared to surrounding normal kidney DNA. The level in normal kidney DNA was basically at the baseline for the assay so by enriching for tumor cells we got the vector copy number to go up significantly over the copy number and surrounding renal tissue that also had circulating blood in it.

We then did insertion site identification via LAM-PCR which I'm sure you are all familiar with. If we looked at this animal, again, we had DNA banked on this animal every six months before the tumor developed. We looked at samples in 2003 and 2004 and we saw a reasonably polyclonal pattern. This is the internal band down here. We saw multiple different bands both in granulocytes and mononuclear cells.

However, when we looked at granulocytes the day before death in this animal, we saw a very, very oligoclonal with these two main bands here that turned out to be real insertions as opposed to some of the junk we have up higher.

And in the tumor itself we saw -- in the enriched tumor scratch we saw the same two bands. If we looked at the kidney that was scratched adjacently but did not contain tumor cells or very low level, we did not see -- we saw some very faint bands in the same region but we did not see these strong bands that we saw in the enriched tumor scratches or in the granulocytes at the same time.

When we did sequencing of these bands this insertion was the chromosome 15 insertion. Now with the entire human genome sequence available we were able to map it to BCL2-A1 which I'll talk about in a minute.

The second integration was not mappable to a specific site. It was quite a short sequence. It had an LTR. We are certain it was a real insertion but it was in a repetitive element so it was impossible to actually map it to a specific site. However, it does not appear to be the chromosome 9 insertion that was pulled out by inverse PCR.

There's two possibilities. Either there were cells, as was pointed out by the CFU analysis earlier that had only the chromosome 15 insertion and then the cells went through another cycle in vitro and got a second insertion in chromosome 9. That's possible. Or that the chromosome 9 insertion is not pulled out very easily by LAM-PCR and we are trying some different enzymes now to look at that possibility.

Then this third insertion or third site may not have been pulled out by inverse PCR which is quite typical. The BCL2A1 locus is shown here. Basically it's a relatively small gene with two exons. The insertion was in the intron and reverse orientation to the direction of transcription.

This is insertion site 1, chromosome 15 specific, PCR using those allele specific primers I talked to you about before and, as you see, basically confirming the data from earlier that over in 2001 and 2002 the contribution from that clone was below the limited detection by this particular PCR analysis. The tumor is very strong as well as in mononuclear cells and granulocytes from the day, again, before the animal died.

So BCL2-related protein A1, another name is BFL1, it belongs to the BCL2 protein family. The whole family is anti- or pro-apoptotic regulators in embryonic development, homeostasis and tumorigenesis. This originally was identified as an early inducible gene that prolongs cell survival. It falls into the anti-apoptotic class of BCL2 family genes, but it also allows some proliferation and differentiation unlike other members of that family.

The problem is it's highly expressed in bone marrow and in peripheral blood progenitor cells even at base line, but it's also been found to be expressed in various primary cancers and related inflammatory tumor infiltrates. It's been reported in gastric, colon, breast, bone, soft tissue sarcoma, and myeloma. It seems to sequester the anti-apoptotic protein Bid to prevent interaction with prop-apoptotic Bak or Bax so in that way it ends up as an anti-apoptotic protein.

When we looked at the insertion site 2 specific PCR, the chromosome 9 site, we also seem to see a marked enhancement of the presence of that clone both in the tumor as well as the mononuclear cells and granulocytes. Again, this was the insert 2 that was seen in the southern blot and by inverse PCR but not picked up originally in our tumor scratches by LAM-PCR.

We looked obviously for helper virus in this animal. We used a consensus PCR that was developed originally to look at the recombinant helper genomes that arose in 1992 in another monkey that was transplanted with help contaminated vector supernatants. Basically we found no evidence for gag-pol sequences present by PCR as a positive control. We are also doing rescue assays as well on serum on this animal but we have not got results of that yet.

Now, in terms of what's the sort of denominator for animals that would be potentially at risk for this type of event. We have a cohort of about 42 rhesus macaques that we've been doing a prolonged follow-up study on. We basically concentrated on animals that had at least stable marking levels at grater than six months of 1 percent or greater. The average level of marking on these animals ranges as high as 35 percent down to 1 percent with a median of around 3 or 4 percent.

The median follow-up now of this cohort that I originally present with Hans Peter Kiem's baboon data and Lochleer therapy last year. Now the median follow-up is 4.9 years. Basically in terms of animals that are out 6 to 7 years like this animal the end is about 7.

In terms of other animals that were transplanted with the same vector that we followed long-term only one. We did some other animals with this vector but the marking levels were so low that we took them off study and released them for use in other protocols.

Ongoing studies. Obviously we're working with Christof von Kalle and others to try to do further tumor cell purification using laser capture and his LAM-PCR basically that can work on single cells. Expression analysis is going to be somewhat difficult because contaminating blood cells already express this protein at reasonably high levels so if we see it in tumor cells I'm not really sure what you actually compare that to. Nonetheless, we are pursuing those studies as we speak in terms of whether -- also looking at whether expression is coming from the vector LTR.

So it's a complicated scenario. The model might be that there was some survival and proliferation advantage for the clone that had the BCL2-A1 or the other hits. That led to hematopoietic dominance early after engraftment. Whether that subclone died out or just became sort of quiescent again is not entirely clear. This resulted in hematopoietic cominance but retained differentiation.

The other interesting thing is if you look at Jan Abkowitz' models in the cat, if you have a low number of stem cells you are much more likely to see these large swings in the percent contribution from individual clones.

As you might expect, if you have only five stem cells contributing, one extra cell division of an early precursor cell could end up leading to a much larger contribution. If you have, you know, 100 stem cells, then it's very unlikely that the behavior of a single cell in terms of the choice of whether to divide or apoptos it's not likely to change the overall marking level.

Perhaps what was seen earlier was actually just to do with the fact that this animal may have had a limiting dose of stem cells, although the animal did engraft on time so I'm not really sure we have a lot of support for that possibility.

We would then say a subclone developed basically at the end of the summer in September that, again, allowed it to expand in the peripheral blood and then perhaps yet another hit leading to overt transformation in the actual myeloid sarcoma tumor.

Clearly there has to have been multiple hits because we know this clone was able to differentiate normally in terms of contributing to granulocytes and other hematopoietic lineages for six years.

However, despite the fact that, you know, it's somewhat difficult due to the lack of individual cells that we can analyze in terms of sorting by flow I believe that the analysis of enriched tumor cell populations would suggest that, yes, the tumor has a vector and, yes, it has this particular clone that had been dominant earlier and, therefore, that it's very likely that the gene transfer in some way precipitated the eventual leukemia in this monkey. What role the drug resistance gene played in this and what role the one cycle of trimetrexate played in this I don't know.

In terms of acknowledgements, I would like to thank Stefania Pittaluga who did all of the hematopothology with us. Bob Donahue and his staff who have taken care of these animals over the years and carried out the experiments with us. The Veterinary Resources Program who did the autopsies and original analysis on the animal tissues. Art Nienhuis and his collaborators at St. Jude who did all the original work with the vector and transplants. Christof von Kalle for molecular analysis.

The people in my lab. Ruth Seggewiss has done all this work with help from Cole Ferguson, Rima Adler, Stephanie Sellers, and Javier Gueniga. Then my long-term collaborator John Tisdale. Susan Wong has been helping us with southern blotting. The Malech Lab has helped us with the universal PCR. Thank you.

DR. RAO: Thank you, Dr. Dunbar.

Is there any specific question? Go ahead, Bruce.

DR. BLAZAR: Cindy, have you ever tried to take bone marrow before the event and add methotrexate in vitro and see if you could select out this code? Or in vivo do you have any samples stored where you could see if the clone is selected out?

DR. DUNBAR: Art? Unfortunately at the time the trimetrexate was given, Dr. Sorrentino's group looked at the enhancement by flow but when they only saw very transient enhancement, no molecular studies were done and I don't believe any marrow was stored from that time.

We have not been doing routine bone marrows on these animals over time just because of the difficulty and labor involved in that. Maybe we should be. There are individual colonies from one year that were stored.

Pat Kelly at Cincinnati is maybe going to look at those again with LAM-PCR instead of inverse PCR but those were not drug selected. Brian and his group and David Williams, another who worked with MGMT and DHFR genes. It's very easy transduce and select in vitro with this vector. Yeah, it does confer drug resistance in vitro.

DR. BLAZAR: It doesn't select out the clone.

DR. DUNBAR: Oh, in terms of selecting out the clone, no, that's a good idea. I mean, no, we have not taken this vector and transduced monkey or human cells and then drug selected and see if we get this particular clone.

We've done and other have done some work looking at what the insertion sites are at the end of transduction. You have so many cells that probably aren't going to engraft and so few true stem cells are cells that are conferred the ability to engraft that I think you're looking for a needle in a haystack

DR. BLAZAR: That's where cells from this particular monkey --

DR. DUNBAR: This particular monkey we do not have any marrow stored. The marrow at the time of autopsy was unfortunately not in any kind of shape that it would actually be viable. The vets did not feel at the time that they were trying to keep the animal alive that doing a bone marrow was feasible unfortunately. We were hoping to keep it alive until we had everything in shape to do all the studies we wanted to do and then it was found dead. That was truly unfortunate.

DR. RAO: Dr. Mulligan.

DR. MULLIGAN: Out of the 42 monkeys that had greater than one percent, what percent had greater than like 10 percent? I think you were rightfully thinking about can you draw any sense about statistics so one out of 42. If you set your bar at 10 percent, for instance, what would be the frequency?

DR. DUNBAR: I would need to go back and look. Stably greater than 5 percent. I would say maybe 20 percent of the animals. One of the animals that is out actually even longer than this animal who was out basically seven to eight years has very high level stable markings, animal RC501 that we've published on extensively.

That animal has stable marking of about 20 percent and it has hundreds of clones that we've identified. So far in our database of 702 retroviral integrations we've not seen BCL2-A1 in any other animals. We've not seen LMO-2. I mean, I just didn't think we'd have time to talk about it today but it may be relevant. We did see 14 insertions out of 702 in a single gene, MDS1/EV-1 complex.

I don't know what Paul is going to talk about Dr. Copeland's work in this area but we find very strong evidence that if you get an insertion in this particular area in this particular gene, it seems to lead to a strong preference for engraftment and long-term presence at a very way, way out of statistically possible frequency. We see 14 insertions in the first entron basically.

None of those animals have developed leukemia. We are following the level of the MDS1 clones in these animals over time and they don't seem to be expanding. It's another interesting finding, especially in conjunction with some of Dr. Copeland's data for it being an immortalization gene.

We also found in our PLoS study that we published that the pattern, as you would expect from a cell line, that the pattern of integration is very different with retroviral vectors compared to any lentiviral vectors and we did not find any single gene that was over-represented by hits with a lentiviral vectors. We did not find MDS1. We did not find in any other genes but especially striking was the difference with MDS1, a finding at 14 out of 702 in the retroviruses and zero out of about 500 in the lenti vectors.

DR. RAO: Thank you, Dr. Dunbar.

Our next speaker is going to be Dr. Dave and while he is setting up maybe we can have Dr. Salomon introduce himself.

DR. SALOMON: Hello, everybody. It's nice to be back. Dan Salomon, Scripps Research Institute and formerly of the BRMAC.

DR. UTPAL: I want to thank the committee for this invitation to come here and discuss our work. The two obvious and vexing questions about the French gene therapy protocol are, (1) why such a high incidence of leukemia, and (2) why the unlikely improbable event that patient No. 4 and patient No. 5 had insertions in the same gene LIM-only 2 or LMO-2.

In the mouse cancer genetics program we've been studying very unique interesting mouse models of cancer. I believe they can provide some insight on these issues. This slide has been reviewed by Carolyn and you guys are well aware of this so I'm going to skip.

These mice are recombinant inbred strains that spontaneously develop leukemia or lymphoma because they have endogenous retroviruses. These are mammalian type C retroviruses that do not encode oncogenes but instead they induce tumors in these mice through the process of insertional mutagenesis. Just by chance the retrovirus may land near an oncogene thereby activating it, or it may land within a tumor suppressor gene and then shutting it off.

The background of these mice also plays a role. As you can see, some strains only develop T-cell, others B-cell, and others myeloid tumors. The point being we don't introduce these viruses. We don't do anything to these mice. We just merely age them.

As they become ill after about six months to one year of age and adenopathy is recognized, we do necropsies, do extensive histopathology to characterize the tumors. Then from the tumor genomic DNA we clone out the insertions. Each tumor in general will have about two to five clonal insertions.

There will be some insertions that are casually unrelated to tumor formation and these are bystanders or passenger insertions. There are other insertions that are repeatedly pulled out in multiple independent tumors and these we define as common insertion sites.

Our definition for a common insertion site is finding at least two independent insertions within a 30 kb window or a 30 kb gene. The probability of finding this as just a random event is on the order of about one out of 100.

If our criteria for a common insertion site was much more stringent and we asked that there were three independent tumors within that same 30 kb window, then using the same probability calculation the chance of that happening as just a simple random event is very rare, on the order of one out of 100,000.

So just some calculations here. The current build of the mouse genome is this. The probability of falling into any 30 kb window is this. Then if one assumes that to take a normal cell and turn it into a malignant cell it requires three specific hits, then the chance of this happening is quite rare in one round of infection. This is only to make the point that tumor formation in these mice is a reiterative event, that you get the hematopoietic stem cells showered with these viruses neonatally.

Some of these viruses may land in oncogenes that result in clonal expansion. Then that expanded population then gets showed again. I think the system actually lends itself very nicely to isolating initiating events, tumor progression events, and isolating cooperating events between oncogenes and tumor suppressors.

All of our data prior to last year was done through inverse PCR. I'm not going to go through the details of this. The viral chromosomal junctions can then just be blasted against the publicly accessible mouse genome and then the precise sites of insertion mapped. These are all publicly available on our website which is the Retroviral Tagged Cancer Gene Database and here is the URL for it.

Ours is not the only model on this website. The other large group that does such insertional mutagenesis is Anton Berns' group in the Netherlands. We have mirrored sites and their data is also accessible on our website. You can search by specific tumor model and specific tumor phenotype.

To date on the website we have 2,000 proviral insertions and 150 of these are common insertion sites and many of these are valid human cancer causing genes. Many of these have no such defined role and that is what we've been trying to work on, trying to define the ones that are novel.

So when we looked at our mouse tumor panels for LMO-2 insertions. To date we have five insertions in the mouse LMO-2 genes. On the top here is the human LMO-2 gene which has two different promoters and in the mouse the structure is similar although the regulation may not be similar. There's also a long transcript and a short transcript. These are five independent tumors with insertions in LMO-2. This is not too surprising because, as I mentioned, many of the common insertion sites are valid human cancer genes and LMO-2 is involved in chromosomal translocations which place LMO-2 under the regulation of TCR delta or TCR alpha.

What was surprising is finding three hits in the common gamma chain gene because there is no previous data prior to this to suggest that common gamma or its deregulation may be oncogenic. What is most striking is that there are two tumors, tumor 98-031 and tumor 7107 which not only have an LMO-2 insertion but they both have common gamma insertion.

This third tumor that has common gamma insertion does not have an LMO-2 insertion but it has an insertion in notch 1 which is also a well described T-cell oncogene. But the probability of this happening is obviously very rare and this implies that there is co-selection or cooperativity between LMO-2 and common gamma-c.

This slide shows that the LMO-2 insertions in every instance in all those five independent tumors are clonal and, thus, initiating events in the formation of these leukemias. The germline bands are shown with the G here and the asterisks show the rearranged bands. In every instance the rearranged band is very strong and in some instances as strong as the germline band.

The right-most southern blot shows it's using a probe to the esotropic envelope and it's just showing that, as has been previously published, these leukemias have anywhere from about two to four clonal insertions.

I should mention that all of these five tumors were lymphoblastic leukemias. they all resulted in large thymuses, large spleens. The histopathology showed that they were T-cell origin. In many cases the T-cell receptor was shown to be rearranged.

As is the case with LMO-2, the insertions in common gamma are also clonal. Again, the rearranged band is quite strong, almost nearly as strong as the germline band. This tumor 7107 was very interesting because using an LM-PCR technique I was able to show that this tumor is clonal for LMO-2 but it's actually by-clonal. There are actually two independent insertions for common gamma-c within this one tumor. They are in different orientations.

This particular restriction digest and probe could not show two bands so two bands are not resolved because these insertions are quite close to each other. Again, this tumor is clonal for the LMO-2 insertion but then common gamma-c was selected for twice. There are two independent insertions for common gamma-c.

Looking at the consequences of this insertion, we analyze this by realtime PCR from whole RNA isolated from these tumors. In the case of LMO-2 as in the case of the human patients, in every instance there was upregulation of LMO-2. LMO-2 expression is upregulated because of retroviral insertion. Interestingly, in the case of common gamma-c there was no upregulation demonstrated. All of these -- both of these results were normalized to T-cells that were isolated from normal spleen.

This particular tumor was an interesting phenotype and it was worthwhile discussing, I thought, because it was difficult to classify. It had germline configuration of BCR and TCR but RT-PCR showed that it had a T-cell specific transcript, the surrogate chain for TCR, this PT-alpha.

And it had expression of two specific cell surface antigens, CD25 and CD44. I think this is very interesting because these two cell surface antigens define a very specific stage in T lymphopoiesis, the double negative two stage where CD44 and CD25 are both positive.

This is precisely the stage where LMO-2 expression is seen and then rapidly turned off. It's as if LMO-2 insertion created deregulation persistent expression of LMO-2 and may have caused a differentiation block so it's as if this tumor was blocked at the double negative two stage of differentiation.

So one other prediction. Carolyn has showed some of the modeling that she had done. One of the predictions that we have also shown is that it's probable that there were so many cells transduced in this gene therapy protocol that he patient may have gotten anywhere from 10 to 100 cells with an LMO-2 insertion. Our window for LMO-2 was much more generous, about 100 kb.

Yet, not every patient has developed leukemia. This implies that there were other hits necessary. This is, indeed, the case. In patient No. 4 there was a 613 translocation seen and in patient No. 5 there was an interstitial deletion that created a SIL-TAL1 fusion transcript. Patient No. 5 was also trisomy for 10.

We went back to our tumor panel modified how we were pulling out retroviral insertion sites and moved to a much more high throughput method using LM-PCR followed by shotgun cloning. Then we tried to clone out as many insertion sites as possible from these five LMO-2 tumors. The results were quite remarkable.

We got a total of from those five tumors 75 integrations cloned. There were 15 integrations cloned per leukemia. Again, about two to five were clonal. The rest were subclonal per leukemia. So the total integrations that are actually common insertion sites in our database were actually 47 percent or 35 out of those 75 meaning they were found in multiple other independent tumors from our database.

There are about seven common insertion sites per leukemia which is quite remarkable because what you would expect is about one out of 75 insertions would be a common insertion site just by looking at how well -- how much of the mouse genome is taken up by common insertion sites. This is what's predicted and, yet, we are seeing 35. It's implying that the insertions in these leukemias are actually causal and they are not just there by chance.

Just within this small subset of leukemias we pulled out common insertion sites besides common gamma chain that also appear to be cooperating with LMO-2. These are the genes that we pulled out in more than one LMO-2 tumor. Again, they are very strong candidates for cooperating hits with LMO-2.

Again, in this tumor I showed you that it's bi-clonal for common gamma. It's bi-clonal for this other insertion IRS-2. Then this is a very interesting gene, MEF2C, which is a MADS box transcription factor. It actually was isolated from three different LMO-2 tumors. It actually is clonal in all three of those tumors. Also in initiating hit along with LMO-2.

It's interesting that every single LMO-2 tumor had at least one insertion in a gene that is involved in a human cancer chromosomal translocation. Many of these cancers were actually hematopoietic. In the case of RAP1GDS this tumor had an insertion in this gene. Then to better validate these insertions that we're finding in the LMO-2 tumors we turn to some microarray data that was done by the Look lab a couple of years ago and it was publicly accessible.

It was microarray of 39 lymphoblastic leukemias. We separated this microarray analysis into those tumors that were LMO-2 high expressing and LMO-2 low expressing. I should say the low expressing many of these tumors were actually negative for LMO-2 expression.

Then we just did a simple t-test to align the genes as to just by p-value and this TIF file just shows the top 50 genes that cluster between LMO-2 high and LMO-2 low. Our first question was what are the genes that are highly expressed in the LMO-2 high group.

Here are just the top 10 from that previous slide. Again, they are enriched in the LMO-2 high group. They are just shown in table form and here is the p-value, their foldover expression in the LMO-2 high group, the genes and their functions.

Our first question was how many of these are common insertion sites. How many of these are in our database. Actually, quite a few and this is really remarkable because it allows our database to help annotate some of this human microarray data.

Then our next question was how many of these genes are found in -- are actually insertions within our LMO-2 leukemias. Actually, in two instances, LAPTM5 and MEF2C these exact genes are not only present in the human microarray but they are actually insertions in the mouse LMO-2 tumors. In the case of MEF2C I actually already mentioned that it's a clonal insertion.

In those instances where we don't hit the exact gene we're hitting in the case like at RAP1B we don't hit RAP1B but we hit it's GTPA exchange factor and its a common insertion site. We don't hit SPINK2 but we hit a paralog. We don't hit STAT5A but common gamma chain is upstream of STAT5A. We didn't hit SYK in the 5 LMO-2 tumors but we hit a substraight of SYK linker for activated B cells.

So just in conclusion these unique models, this mouse retroviral insertional mutagenesis, they are a very powerful cancer gene discovery tool. I've just shown some data with comparison with the microarray that they can very accurately model human disease.

With respect to LMO-2 and common gamma this data suggest that they are cooperating to induce leukemia because there is a very low probability that both genes are hit in the same mouse tumor by just a random event.

In the case of the gene therapy cases one hit is delivered by the retroviral insertion and the second hit is delivered by transgene expression. On the whole we think this is actually good news for gene therapy because if one looks at our database and clears genes that are represented in the database, not every gene is going to be oncogenic when expressed in a retroviral backbone.

I'll just stop there and take questions.

DR. RAO: Go ahead.

DR. WOLFF: When you went back and looked in LM-PRC at your tumors and you found so many more integrations, would you suggest that these tumors were oligoclonal and that many of these integrations were only found in some of the tumor cells?

DR. UTPAL: Yes, that's true. That's true. Each tumor had about two to five that were probably clonal and then the rest were subclonal insertions present in the subpopulation of cells.

DR. WOLFF: How do you explain what their role might be in the development of the tumor itself if they are only in some of the cells?

DR. UTPAL: The data that suggest that they may actually be causal is the fact that they are pulled out in multiple independent tumors besides the LMO-2 tumors. It's true that -- it's possible that some that I've shown in the case of common gamma-c and MEF2C they are clonal along with LMO-2 meaning they are initiating events. But the others could be tumor progression events.

DR. LEONARD: Related to mechanism you show that the LMO-2 integration are associated with the elevated expression of LMO-2 but for the gamma chain you said that they were not associated with elevated expression. Is there any evidence in the tumors of augmented gamma chain function either JAK-3 activation or higher response to any of the cytokines? Had you meshed the observation, the integration but yet no change in the level of the gamma chain protein?

DR. UTPAL: I wish we had preserved the tumors in a way we could analyze that but we have not. It's possible that common gamma was upregulated at some point but at the time the RNA was analyzed at the time that these became frank tumors that we were not seeing over expression. That's one possibility that may explain that we're not see over expression.

There are instances of insertions in MIC, there are insertions in ETS1, and there are chromosomal translocations of LMO-2 where there's deregulation and not over expression of these oncogenes. Including there's a chromosomal translocation of LMO-2 in a patient that shows no change in expression level.

That would imply that shutting off common gamma at some point in development is important and that retroviral insertion is eliminating that regulation. That's just another hypothesis, another possibility to explain this.

DR. RAO: Thank you, Dr. Dave. Our next speaker is Dr. Christopher Baum from Hannover Medical School.

DR. BAUM: I would like to thank the committee for the invitation. The problem with the slides is due to the facts that I have the same initials as Claudio Bordignon. This work that I'm going to present you is primarily performed by Zhixiong Li and Ute Modlich for discerning and conducting the mouse studies. We had help by histopathologist Kenji Kamino and Brigitte Schlegelberger for cytogenetics.

The molecular analysis of the integration sites was a collaborative effort with the lab of Christof von Kalle, especially Manfred Schmidt. And also Boris Fehse and Olga Kustikova who were at the same time tested at MPCR in comparison with LAM-PCR. Actually we found both methods to be equally valid when looking at the very dominant integration sites.

I just wanted to have this for the introduction to remind us that insertion mutagenesis is one but not the only side effect of retroviral transgene technology. It has to be put in context with some other issues especially when we culture cells only for very short term before infusing there might be persistence of infectious particles in vivo.

There is the problem of mutations and recombinations and reverse transcription which is true for any type of retrovirus or lentivirus and the notorious issue of ectopic transgene expression levels on physiologic regulation of transgene expression. All this has to be viewed in the context of the cell type that we are engineering and also the milieu which is very disease specific as you all know.

Many people would state that the side effects seen in the SCID-X1 trial have not been predicted by mouse vector studies. While to some extent there was a predictive study shortly before the side effects were reported. Published a short paper where we found a very similar constellation in a mouse gene marking study.

In that case a single retroviral vector integran had strongly activated EVI1. We've heard about EVI1 before. It's a transcription factor that by itself might immortalize mouse hematopoietic cells but is not sufficient to cause leukemia so we asked what might have been a contributing factor here. I read a paper where it has been shown that the transgene can cause signaling side effects in a very context dependant fashion.

This is the truncated lower finishing EGF-R factor receptor showing this study to enhance signaling through receptor choice in kinase TrkA when being triggered by nerve growth factor. It's still a cytokine-dependent mode of action which can lead to a growth advantage. Importantly the same transgene expressed from retroviral vectors in T-cells did not cause any side effects in human studies.

This is a similar constellation. You've seen that so frequently. Again, I just would like to point out we have a transcription factor activated and the transgene that is not physiologically regulated and still needs cytokine signalling to give a survival advantage so it's not an autonomous signaling situation and whether or not this combination of receptor gamma chain and LMO-2 chance or necessity is the subject of an intense debate.

We have contributed to that but I think the data shown by Utpal Dave with quite strong evidence that we need really better experimental systems now to look in the vector context how exactly this collaboration takes place. It really seems to be a collaboration going on.

Others have contributed important work by looking in unbiased cell systems on the insertion profile of different retroviral and lentiviral vectors. We learned from that there is, indeed, a difference depending on the type of retrovirus and lentivirus used. But on a more global perspective when you look at these data in unselected cells, the differences are statistically significant but do appear to some extent also minor.

The probability for HIV to be within ref seq is two-fold higher than for MLV. And to be plus minus 5 kb after transcription starts is only two-fold lower so there's no absolute distinction between different vector families. Therefore, we need function and models to evaluate the toxic side effects.

We have a proposal published in Blood early in 2003 that, therefore, we have to change paradigms little bit in retroviral vector gene therapy while still going for improved technologies and strategies. Also, provoke preclinical worse-case scenarios for those findings. This proposal hasn't found so many friends initially but I think it's helpful and I would like to convince you actually in the few remaining minutes.

This is just a theoretical consideration together with both Fehse and a statistician. We thought, okay, if a cell has no entry limitation to a retroviral or lentiviral transduction, what is the probability of accumulating more than one insertion? If you have a mean copy number of two for a cell formula with, say, this result in about 87 percent positive population and then based on Poisson statistics more than 40 percent of the population would have a more than three and up to seven insertions.

Now, we know from the work with replication competent retrovirus the simultaneous presence of two or more oncogene insertions gives rise to leukemias. Can that still be transferred to a vector situation where we have a very limited number of insertions.

So we designed the following mouse model going into a normal mouse that has no increased risk of hematologic malignancies C57 plus 6. We designed a protocol that allows precise dose escalation which was published in Experimental Hematology. It doesn't use any cytotoxic drug to mobilize the cells. We have lineage depletion of cells to have a mixed population of progenitors and stem cells.

Then they are put in serum-free cultural conditions and then the virus can be used at high or low MRI and the cells can be cultured for short period after transduction or expounded in vitro prior to transplantation. Then we would go through a cellular bone marrow transplantation to have an additional challenge in the system that kind of triggers replication of stem cells.

This is prognostic in the sense that we look actively for signs of leukemia and lymphoma. We have a histopathologist evaluating the samples and, of course, go for transgene status. The study just published in Blood online shows the first data obtained in this model where we compare the vector that has a retroresin protein and the vector expressing multi-drug resistance.

Why MDL-1? That is because a previous paper has shown that mice develop leukemias when they are challenged with high vector doses of this multi-drug resistance gene vector and we wanted to ask whether that is because there are multiple copies and geneotoxicity or whether that is due to the high expression of MDL-1.

The details can be read in the Blood paper. Just briefly bottom line very easy. Dose escalation is leukemogenic with vectors in mice. There is quite a high incidence of leukemia especially when you explant the cells in vitro prior to transplantation and then the incidence is up to one in a million treated cells so something that can be put then maybe into large scale studies.

In the DsRed control group we had a similar incident when cells were not expanded prior to transplantation. We didn't test in this study what would happen if we would expand them prior to transplantation with this vector.

In contrast also quite valuable information from this study prospectively analyzed mice that had low-dose gene transfer of any of these vectors did not develop leukemias. The incidence must be below 1 in 28 treated cells.

The pathologist told us that the leukemias had quite a valuable phenotype. There were verified cases myeloid without and with maturation and also T lymphoblastic arguing that the genetics behind the leukemias was more stochastic.

Finally that was confirmed by looking to the integration sites. And without going into any detail, it really looks pretty similar to what you find with replicating retrovirals so the selection goes then for cells that have similar constellations and kind of reminiscent of the complications that we had in the earlier studies where only one vector integrant was found in leukemia.

We have always a combination of transcription factors shown in green or nuclear regulatory proteins shown in green. Shown in this orange color growth factor receptors or molecules acting downstream of growth factor receptors such as Sos1 which would activate Ras.

Here this is G-CSF receptor, for instance, that is being hit. This seems to be a dangerous constellation, disregulated growth factor receptor signaling plus transcription factor which is compatible with current concepts of leukemogenesis.

When you have leukemias with very high integration sites scenarios you also find increasingly a kind of innocent bystander mutations or insertions in other genes that are not predicted to have the signaling function.

Some statistics can be derived although the sample size is still limited but it looks again pretty similar to what we've seen from Utpal Dave's RCR studies. There is the probability up to 40 percent to proto-oncogene when you look only into the tumors so that a similar selection is going on as in the RCR studies.

I want to point out one important thing here shown in red. Even the healthy animals that were sequenced as controlled recipients had an over representation of proto-oncogene hits although not as strongly as the leukemic animals.

This is more than randomly expected. We though, okay, there might not be more than two percent of the genome filled with proto-oncogenes so we still have at least 5 percent over representation of such hits in healthy animals.

In a separate study, which I don't have the time to go through in any detail, we can then formally show when we take those clones for cell transplantation that they dominate hematopoietic in cellularly transplanted recipients, however, without causing leukemia.

That also tells us the single proto-oncogene does not necessarily give rise to leukemia. It can only give rise to clonal dominance. We have some evidence by local specific PCR that some of those clones die out. When cellular transplanted they are extinguished so sometimes it also causes selective disadvantage in cellular transplant situations.

So this leads to this simple scenario where we would say when we have a situation with a single activated oncogene it might well give rise to transient cell expansion. But in case there's no independent hits occurring at that time or during the survival of this clone, we would predict there is a low risk of malignant progression.

Actually, there might sometimes be a therapeutic advantage of having expanded still healthy cells by the fact of insertion -- by the effects of insertion mutagenesis. So benign dominance or extinction of such clones is the most likely outcome. In scenario B which would be a high-risk scenario for gene therapy, you will have a situation where right from the start of the existence of the clone there are two hits.

At least one activated oncogene and side effects of transgene expansion which we term phenotoxicity in this blood review or additional genotoxicity by another independent proto-oncogene insertion or even two or more proto-oncogene insertions. And then the probability of expanding and really transforming the cycle is strongly increased resulting in a clear overt risk of malignant progression, even in clinical situations.

Just in brief summary, this worse-case approach seems to work, although the data is still early and preliminary to some extent so we can't generalize too much. We would say the escalation in the mouse model can be proposed for safety assessment and in a normal mouse model that has no endogenous retroviruses being activated.

This approach still needs to be adapted for disease-specific settings and that is a real challenge and I hope that people are working on that. We are currently not doing this work.

The single proto-oncogene hit is not sufficient for leukemia induction not even in the mouse which is more easy to transform than the human cell background. We need collaborative efforts such as the database shown by Utpal Dave from the Copeland Lab and also maybe some appointments of preclinical standards which would be quite helpful.

Keep in mind that the vector type which it's gamma retrovirus or lentivirus or whatever, it's just one out of many important variables. We have to consider exactly the type of cyst elements and the architecture of the vector, the type of CDnase being present, cell type and milieu factors.

Just three short outlooks. First, we are organizing a retreat which will take place on the morning of the opening day of the American Society of Gene Therapy meeting together with Christof von Kalle and David Williams. If you are interested in further details, send an e-mail to Keisha.Steward@CCHMC.org.

One more comment regarding insertion of distribution. Hits can be found everywhere not only in this famous 5 kb window which is the preferential target. For MLV you find them sometimes downstream and sometimes far upstream of a gene of interest so we can't really say how other retroviruses will decrease or increase the risk based on their insertion pattern.

We can modify vectors by taking out, of course, the enhancer promoter from the LTR. This really seems to be mandatory for some of the stem cell applications we are planning. This can be done also in the gamma retroviral context.

When we improve on processing of the vector, we might be able to work with enhancer promoters internally. Then we could also test whether the LTR is still useful by incorporating these famous insulator or scaffold-attachment regions preventing interactions with the neighboring allele.

With this I would like to stop and thank you.

DR. RAO: If there are no specific questions, thank you.

Our next speaker will talk on use of retroviral vector mediated gene transfer in ADA-SCID clinical trials. Dr. Donald Kohn.

DR. KOHN: Thank you. I'd like to thank CBER and the committee for inviting me to speak. I'm going to just give a little background on ADA deficiency as another cause of SKID, talk about some of the alternative treatments which are PEG-ADA enzyme replacement therapy, and for kids lacking a matched donor haploidentical bone marrow transplant results, and then briefly summarize the results of our studies and Dr. Bordignon studies that you've also heard about.

So ADA deficiency is responsible for about 20 percent of cases of human SCID and is fatal if untreated due to usually opportunistic infections. It's somewhat of a pioneer that it was the first foremost SCID where the genetic cause was identified somewhat by accident in 1972. The first time the responsible gene was cloned which is why it was probably the first gene for which gene therapy was attempted in SCID. Gene therapy was first approached in 1990.

As I'll talk about, it's a form of SCID where there is one other therapy other than bone marrow transplant that is effective and that's PEG-ADA enzyme replacement therapy which was approved as an orphan drug in 1990.

This drug is given by bi-weekly typically intramuscular injections and needs to be given on an ongoing basis. If therapy is stopped, immunity wanes. It can restore and partially sustain immunity and I'll show you some data on that. And it's a very expensive drug in the range of $250,000 to $500,000 per patient per year ongoing which gets to be a problem. Now as the first cohort are starting to reach adulthood insurance coverage is an issue.

The pathway. It's the enzyme that basically deaminates the amino group off adenosine or deoxyadenosine to inosine which is salvaged. T-cells have very high level of kinases that phosphorylate and trap. Especially dATP is one of the metabolic toxins that probably poisons off the lymphoid cells. In the absence of ADA the lymphoid cells are particularly affected in its paneling for TBNK.

Enzyme replacement with PEG-ADA can correct the metabolic abnormalities which is usually monitored by looking at red blood cell levels of deoxynucleotide metabolites. It has a variable restoration of immune functions with approximately 20 percent of patients not responding and about half not developing adequate B-cell function so they remain on intravenous gamma globulin.

The last large survey is by Mike Hershfield showed the overall survival of patients with ADA-deficient SCID treated with PEG-ADA was 83 percent of the 113 patients he ascertained. 73 percent if you include the patients who underwent bone marrow transplants. As I'll show you, the results of bone marrow transplant for this disease are not good.

Ten percent of the patients developed neutralizing antibodies and there have been autoimmune syndrome seen in five patients, three of which were fatal.

These are some unpublished data. Actually a medical student worked with me over the last two years has done a retrospective chart review of nine patients who we follow in our immunology lab looking at their immune function on PEG-ADA. This is a graph showing their CD3 counts over time prior to beginning PEG-ADA so basically at the time of diagnosis of SCID they tended to have T-cell counts under 100. One who presented at six years of age had a slighter higher T-cell count.

The maximum they reached was typically about one year after treatment which still was less than 5 percent of the normal range of age-corrected CD3 cells. It does not normalize their immunity. In fact, over time all of them show a waning of T-cell numbers. The most recent time looked at which was now approximately two years ago for this retrospective study was again back down under 500 in every case and some are quite low.

While it gives some partial restoration of immunity and these patients are alive for the number of years indicated here on PEG-ADA 4211 it does have limitations in terms of immune efficacy.

So the other treatment is bone marrow transplant for ADA-SCID and for children with ADA-deficient SCID who have a match sibling, the results are fairly good as they are for other forms of SCID where there is a match. There aren't large series of ADA-deficient SCID transplants reported but generally transplant results are 75 to 90 percent.

There have been some neurologic and behavioral alterations observed in long-term follow-up studies. Again, this is largely anecdotal. The majority of patients don't have a match sibling and, therefore, are faced with a non-HLA identical or non-sibling identical transplant.

Dr. Buckley presented at the European Society for Immune Deficiencies in 2002, the experience at Duke which was 33 percent engraftment from an NF5. Then a paper published in Lancet from the European Registry showed that overall survival for patients undergoing haploid transplants in Europe was 23 percent. The next slide I'll show you results for other SCIDs to contrast that.

Then Adrian Thrasher presented data also at the European Bone Marrow Transplant meeting for ADA-deficient SCIDs of 13 that have been transplanted at Great Ormondt Street. There's 84 percent survival with a matched sibling, 50 percent with a matched unrelated donor or cord blood, and 23 percent with haploidentical transplants.

These numbers are shown here again from the European experience, 88 deficient SCIDs undergoing matched unrelated or haploidentical transplants, in other words, not sibling, have a 23 percent survival rate whereas other forms of SCID categorized as T-B+ which would include the X-SCIDs is a 50 to 66 percent change so ADA-deficient SCID tends to do worse without a matched sibling transplant.

So the gene therapy that's been done there are a number of trials. I think I've gathered all the numbers together. The original T-cell trials done by Mike Blaese and French Anderson at the NIH in 1990 and by Claudio Bordignon were reported for a total of eight patients.

Then in the early literature which is the early '90s a total of eight patients were described in which bone marrow CD34 cells were the target. Basically to summarize all of these studies, the marking was relatively low and there is no evidence of any clinical benefit.

So in that early era we did out first study in collaboration with Mike Blaese at the NIH treating three newborn infants with cord blood CD34 as the target. CD34 cells were isolated and transduced with the LASN retroviral vector using a combination of cytokines that we now recognize are not very good at getting early stem cells into cycle. The cells were given back on the fourth day of life with no conditioning and the patients were started on PEG-ADA as the standard of care.

This is data that we published a number of years ago showing the gene marking in one of the patients. The circles are peripheral blood mononuclear cells and over the first two or three years marking in those rows to about the one to three percent level, whereas marking in myeloid cells remained low at about .1 to .01 percent or one in 10,000 roughly.

This is thought to represent some selective accumulation of T-cells and samples that were showed that, in fact, the T-cell marking was higher than, for example, as compared to monocytes taken out of the same sample. There was some degree of selective accumulation but never enough to really have any convincing effect on immunity.

This patient has also been studied in collaboration with Christof von Kalle, a paper published two years ago on LAM-PCR. We found a surprising finding in that there was essentially oligoclonal gene marking in the patient over time. This is the same patient that I just showed you, the PCR marking quantification.

This is LAM-PCR showing that there was a predominant integrant that was the majority of the gene marking over eight years of follow-up in both the PBMCs and the T-cells but not so much in the myeloid cells. There were other markers seen. Again, as summarized in the study, LAM-PCR revealed the stable presence of a predominant vector integrant in TM myeloid cells over the past eight years.

T-cell clones grown from the peripheral blood eight years after the neonatal gene transfer indicated that a single pre-thymic stem or progenitor cell accounted for the majority of gene marking and polyclonal T-cell production.

In contrast to the sort of monoclonal malignant outgrowth of leukemia you've seen in the other patients we believe this represents just low-level marking basically at an almost monoclonal level but not in anything that has resulted in transformation.

In fact, this patient is still marked at the same levels. This is QPCR that was done last month showing peripheral blood mononuclear cells continue to be marked at about 3 percent whereas myeloid cells are still marked at one in 10,000. There has been stable persistence of gene marked cells in this patient, although at a low level over this 11 year period now.

So based on the findings from this first study we developed a second study, we being my group at Children's Hospital, with Claudio Bordignon -- I'm sorry, with Fabio Candotti, Cindy Dunbar, John Tisdale at the NIH. The study was basically intended primarily to take care of -- to apply advances in the field which was to use an gibbon ape leukemia virus pseudotype for which there was some evidence maybe more effective than the amphotropic envelope and to use a combination of cytokines that had better evidence for helping stimulate proliferation of early stem and progenitor cells on recomitant retronectin during serum-free.

So the study of IND application was made five years ago in 1999. It was approved about two years later. Four patients were enrolled between August of 2000 and January of 2002 and their characteristics are shown here. These were all patients who were on PEG-ADA at the time so they were somewhat older.

Two were 15 and 20, two were sort of a slightly younger cohort of five and four. These were their CD34 cell doses, lower than typically given in the X-SCID studies. This was just the level of gene marking in bone marrow colonies grown from the cells prior to transplant.

This is the gene marking in the four patients looking over the two years of proposed follow-up. The two older patients shown on the left show marking that really only was evident in the first three to six months and then became undetectable beyond the one-year period in both patients.

The two younger patients show persistent gene marking at the level of about one cell in 10,000 over the period and continue to have that in the long-term follow-up period. These are somewhat disappointing results and you can contrast these with the results that Dr. Bordignon showed in the opening comments where they have marking in their myeloid cells one in 10 and the lymphocytes basically 100 percent.

He talked about his trial just to recap a couple of the points of their study. They reported the first two patients in a science paper in 2002 on ADA-deficient SCID patients and the major advances they took were to give some chemotherapy, busulfan at 4 milligrams per kilo is about a quarter to an eight of what would be an ablative dose in a full transplant setting to make space, so-called non-myeloablative conditioning.

The patients were not treated with PEG-ADA which could blunt the selective advantage of the corrected cells. In fact, they reported there was immune reconstitution within six months. The T-cells were marked at the 10 percent level and the myeloid cells, which would not have an advantage from having the gene, were at the 7 to 12 percent mark.

As he mentioned, since that time they have treated four more patients with a good immune recovery in all of them. These are just some of the data that he showed you on the patients, their ages, their cell dose given. Then one last bit of data slide and that's just to show the immune reconstitution from the Italian study compared to the French X-SCID study.

This was from data they published in a science paper when they reported their first five patients in X-SCID compared to the six patients now -- or, I guess, the first five patients treated in Milan. You can see that there is a lower level of lymphocyte recovery than in the X-SCIDs and whether this represents less of an overshoot or just a lower cell dose is not clear.

So getting back to our study, so when the first cases of X-SCID were reported in September of 2002, we were placed on clinical hold following our initial response to the FDA in terms of modifying the consent and limiting the age and cell dose. The hold was lifted finally just in December of '03.

We then spent a year having an amendment protocol modification reviewed to include giving busulfan and withdrawing PEG-ADA, as well as with the age and cell dose limits that were agreed upon to get off hold at that point. This was then fully approved finally just in January of '05. We were ready to begin patient enrollment when the third case was reported and we were placed back on hold.

So then just to summarize for ADA-deficient SCID PEG-ADA enzyme replacement therapy is palliative but immune function is significantly below normal. There is a poor outcome with haploidentical bone marrow transplant.

There have been no adverse events related to the gene therapy procedure in at least 18 subjects, some of whom, but not all, have had retroviral transfer gene presence for more than 10 years. There has been a very good outcome from the gene therapy studies in the Milan group using busulfan and no PEG-ADA. Thank you.

DR. RAO: Thank you. Any specific questions? Why don't we take a short 10-minute break and then get back to the questions and deal with them.

(Whereupon, at 2:11 p.m. off the record until 2:23 p.m.)

DR. RAO: So clearly all of you heard a lot of data and I'll just remind everyone of the questions that Carolyn posed right at the beginning. I'm actually going to read out the first question. "The Code of Federal Regulations, CFR Part 312.42 defines the basis for FDA to place a study on clinical hold. CFR 312.42(b)(IV), insufficient information, was cited previously as a basis for placing INDs on clinical hold on response to the development of leukemia in subjects of X-SCID clinical trials.

However, we note that the CFR 312.42(b)(I) that states FDA may place a study on hold if it finds that human subjects are, or would be, exposed to an unreasonable and significant risk of illness or injury. With this requirement in mind, please discuss the current incidents of leukemia (3 of 12)."

It's important to remember that in the last six months it's an increase one. "And death of one subject from leukemia reported in the clinical trial in France relative to the potential benefit of retrovirla vector-mediated gene transfer in X-SCID patients.

Consider in your discussion (1) the risk benefit issues for gene therapy versus haploidentical bone marrow transplantation, and (b) the incidence of leukemia associated with retroviral vector administration that would make clinical trials of this therapy unacceptable in X-SCID and would this advice differ for a subject in another clinical trial for leukemia? Would another subject that is due to leukemia influence your recommendations?"

The questions are really quite specific. Before we go to that, I'm going to really try and ask Dr. Buckley to make a comment for the first section of the question and ask her to lead off the discussion.

DR. BUCKLEY: Let me just comment first about the question that Carolyn brought up earlier which was what is the incidence of leukemia in SCID patients who were transplanted. It's not really leukemia that they have developed. It is more of a lymphoma or lymphoproliferative type of problem.

In the presentation I gave two and a half years ago I cited Elaine Jaffe's article where she had summarized the cases that had been published. In our own patient population at Duke where we transplanted 141 SCIDs over the last 23 years, we've had four patients who have had EBV lymphoproliferative disease but we've not had any who developed clonal proliferations like leukemia so I'm not aware of leukemia itself being a problem in untransplanted or transplanted SCIDs.

DR. RAO: And the lymphoproliferative disorder needed to be treated with steroids and that was enough?

DR. BUCKLEY: Well, it's usually fatal. I mean, we have more recently used Rituxan to get rid of the house for the EBV but EBV can bind to epithelial cells and other cells in the body and it doesn't completely work.

DR. HIGH: Well, I wanted to ask as a point of clarification to Dr. Buckley sometimes in the setting of bone marrow transplantation there can eventually after a very long period of follow-up be a phenomena where the bone marrow seems to exhaust. I was wondering what is the duration of follow-up for bone marrow transplantation in X-link SCID and are there data that speak to that point?

DR. BUCKLEY: Well, probably the largest series are those from Europe and I would imagine the data we have from our place. We've been following them now for 23 years in May. We have patients who are in their early 20s who graduated from college who are doing very, very well. But each patient is different.

In large part it depends on when they were transplanted because if they were transplanted after they developed CMV or adenovirus or some of the other viruses that you can't really treat, then they had a more difficult course and the degree of immune reconstitution was not perfect in those patients.

Whereas those that we transplanted in the newborn period, or actually in the first three and a half months of life, we've only lost two patients out of 39 that we transplanted in the first three and a half months.

One thing I guess I should mention, we do not use any pretransplant chemotherapy for any of these patients. It's very effective. Of those 39 we only had five that had HLA identical donors so 34 received T-cell-depleted half match mother or father bone marrow.

One of the approaches to try to improve treatment for SCID would be to make a diagnosis at birth where you could then treat shortly after birth and then you would have a better outcome. But in terms of long-term follow-up we have X-link SCIDs who are in college or who have graduated from college who are doing very well.

DR. HIGH: Well, it sounds also then as if the survival in bone marrow transplant patients who with a haploidentical donor that you might give better numbers for that now than in the 117 patients that were reported on in the slide that we saw from Carolyn Wilson early?

DR. BUCKLEY: Well, are you talking about from our place?

DR. HIGH: Well, yes. But you probably have the best numbers anywhere, right?

DR. BUCKLEY: Right. The overall survival of the haploidenticals is 74 percent and that's over 23 years.

DR. HIGH: And that's at your place?

DR. BUCKLEY: Right.

DR. HIGH: But if you --

DR. BUCKLEY: Well, I summarize that in the slides I gave two and a half years ago where in the published reports of patients who have been treated at a variety of different centers with a variety of different methods.

I think in the European series probably about half, or maybe a little bit more, do get pre-transplant chemotherapy. The overall survival of the half match and the ones who had been given pretransplant chemotherapy was something like about 41 percent.

So I think knowing that these patients come in with infections that are very difficult to treat and then if they are given chemotherapy, that takes away other host defense modalities and maybe that accounts for the poor outcome.

The importance of long-term follow-up is really very high and I think the European group does have long-term follow-up on theirs and we are going to be putting together our long-term follow-up very soon.

DR. RAO: Bruce.

DR. BLAZAR: I wanted to ask for both ADA-deficient SCIDs and non-ADA-deficient SCIDs has the outcome data changed more recently and are there strategies that we should be discussing or being considered that will change the field in the next few years. And also can you make any comment on cord blood transplants in general? The numbers here are quite small.

DR. BUCKLEY: We have not done cord blood transplants in our SCID patients except for those two or three that we did mainly because the cord blood protocols call for pretransplant chemotherapy. Not only that but they call for cyclosporine twice as long after a cord blood transplant as it is after a regular transplant. By taking out the T-cells we don't have to give any post-transplant GBH prophylaxis either so that's why we've avoided cord blood transplants.

But getting back to your original question, the thing that occurred to me when we were hearing the different presentations is that I think we have to think about SCID now as 10 different diseases because there are 10 different genes that when mutate give you SCID. Maybe we shouldn't consider them all to be the same. Maybe having adverse events with X-link SCID may not have any bearing on what happens with ADA-deficient SCID.

Your other question was what about the recent outcomes for transplants in ADA-deficients. It really hasn't changed since the presentation I gave last time and that is that the ADA deficient SCIDs do have the poorest rate of engraftment and that is because if they received a blood transfusion or anything that would empower their own T-cells, then they are more prone to reject the graph.

Resistance to engraftment is a bigger problem with ADA deficient SCID than it is with any of the other types of SCID with the possible exception of RAG-1 or RAG-2 or Artemis.

DR. BLAZAR: So basically the baseline data that we heard about is really the current state of the art and probably in the near future would be used as the statistical baseline to determine how gene therapy compares.

DR. BUCKLEY: Yes, but I think you may have to look at each one of these conditions as a different disease.

DR. SALOMON: So I think that's an interesting premise that you would have to look at each of these as a different disease and it kind of brings up something that has come up here that I think we should discuss for a minute and that is the premise that we didn't have to consider it as a different disease was based on the thinking that the insertion was a function of the vector.

It was postulated in addition that there was an exacerbating factor in that the children were very young which is kind of interesting in that this third case is one of the older children so so much for that one. If it was dependent on the vector and was not associated with the therapeutic gene, the transgene, the common gamma chain, then it wouldn't really matter.

That was why we thought two years ago when we reviewed this last. Now I look at some of this data and I think it gets back to the point you're making is if one believes that the therapeutic transgene was part of the hit, then it's not just the vector's insertional mutagenesis capacity which would be true for all SCIDs but now it becomes a factor of the disease so that, I think, is really an important thing to think about.

The last comment I would make is I find it very interesting that the common gamma chain came up multiple times in the gene expression array and analysis. That was beautiful work.

DR. RAO: Dr. Mulligan.

DR. MULLIGAN: Well, I think this is a great question. I think it's a question we will be able to sink our teeth into. I don't think actually when we get to the technical things that I actually know something about that actually will get very far. I would say the most important person that I'm listening to is Dr. Buckley.

I think it's a question if we -- I'm sure you don't want to know too much about retroviruses but if we tell you essentially it's our wisdom that another patient or two or maybe six or maybe seven out of these 10 will get leukemia, and you're the physician that's weighing the risk benefit, what would your answer be?

Let's just say that the answer is about 60 percent and about 60 percent will get leukemia within five years or so. The treatment -- you would know better the treatment for leukemia there's maybe variable experience with but you would probably have some assessment.

I think what everyone should want to hear is based on setting it at whether it's 50 percent, 60 percent, 70 percent what do you think is the risk benefit. Is it worthy to continue. The other sideline that I think is important is that what difference does it make that you get good results, better results than other people at doing nongene therapy treatments.

I know it's a hard thing to think of but if you have a group that wants to do gene therapy and their capacity to actually do conventional bone marrow transplantation is not so hot; that is, their results will not be so good, what kind of dilemma does that leave us with?

DR. BUCKLEY: Well, it's going to be difficult for me to answer the last question that you just posed because at different centers different people are in charge. It depends on whether someone who is a hematologist oncologist is doing the transplant or whether there is an immunologist involved because if it's just somebody who does leukemia patients or other types of patients for bone marrow transplants, they tend to treat the SCIDs the same way so they all are put into the same boiler plate.

Then I think that is where you see the outcome not nearly as good. Getting back to what I would do if I were a physician and I had an X-link SCID I wouldn't consider gene therapy right at this point because I think that we are still evolving what the future might be for these children that were in the first trial.

The other reason I wouldn't consider it is that of all the different types of defects the X-link SCIDs are the easiest ones to get a T-cell depleted half-match transplant into if they don't come in with ednovirus or EBV or CMV or some other serious infection on board.

DR. MULLIGAN: Can I just press the question? Would you say that it would be an unacceptable risk? The risk would be significant such that this would not be a treatment that you would offer?

DR. BUCKLEY: For that particular defect I would not.

DR. MULLIGAN: Okay.

DR. RAO: Go ahead, Bruce.

DR. BLAZAR: Then to extend that you had a caveat that if they came in with infection, etc., is there any circumstance under which you would favor gene therapy over haploidentical transplant any patient characteristics that one might segregate the risk into a much higher risk versus a lower risk?

DR. BUCKLEY: No. I mean, if you're talking about X-link SCID I don't think there would be any patient characteristic that would affect that.

DR. BLAZAR: Active infections, number of infections none.

DR. BUCKLEY: No, because if you look at the first paper from France where they show the kinetics of immune reconstitution, for the first five patients that they did it showed that it took 90 to 120 days for the autologous gene-corrected T-cells to develop and become functional. That is the same kinetics you get with the T-cell depleted half-match.

MS. BALLARD: What about after failed repeated bone marrow transplants?

DR. BUCKLEY: Well, that is a good point. I think in that circumstance there would be an exception.

DR. RAO: So if you look at the earlier hold that they had, one of the choices was to consider people who had failed bone marrow transplants. Would that be still a reasonable criteria?

DR. BUCKLEY: Yes.

DR. LEONARD: I guess the other relatively rare category would be someone who does not have the possibility of even haploidentical donor due to death of both parents or something like that prior to diagnosis, adoption, whatever. That presumably would be another category

DR. BUCKLEY: Well, we've not encountered that so far but presumably it would be, yes.

DR. RAO: So would that be -- I mean, a general statement in comparison would be if you don't have an haploidentical donor there isn't any other specific risk, is it in the same category as cord blood transplant in your mind for X-SCID specifically? Then Dr. Mulligan.

DR. BUCKLEY: I think I would probably put it in the same category, yes.

DR. MULLIGAN: First, another question for Dr. Buckley. Before the first patient got leukemia, and maybe even before they treated the first patient for X-link SCID, what was your point of view about the rationale for gene therapy for X-link SCID? That is, has your opinion shifted significantly because of these toxicities or were you lukewarm to the gene therapy on the general principle that there were alternative treatments?

DR. BUCKLEY: Well, I must say that I was like everyone else, thrilled that it had worked so well. We were hoping to be able to do that in our patients but after the first series of adverse events occurred, then I had a different opinion.

DR. MULLIGAN: The second point I make is that if you walk down this path, one of the difficulties that we'll have to deal with is that how would it be the case that one would ever return to this disease with gene therapy? That is, what would be the technical fixes that we would really have to have in place before you would come back to treat this disease?

The worry that I would have is that unless people begin to think about those sorts of things, then you are going to wipe off a number of diseases off the list and never really gain the clinical experience to see whether things helped.

What I said earlier about we're not going to get anywhere with all the technical things, I think these are going to be things that everyone is going to encourage people to do but no one is going to -- they are not at a point technically where you could legislate that you must use this kind of suicide marker, etc.

But presumably the value of encouraging people to use these features is they can test them in people and hopefully get a sense that there is something good happening. If you don't have a population, if you're not using that population, I wonder how you'll ever get to that population.

DR. BUCKLEY: I think the thing that would make me want to try it again would be if you knew that you could direct the site of insertion and that basically is what you're saying.

DR. RAO: Bruce and then Dr. Murray.

DR. BLAZAR: I wanted to return to the point of failed bone marrow transplant as an indication. There is still the option of T-depleted matched unrelated donor which biologically might not be so different than haplo and it doesn't seem like -- there is some data from the European registries to suggest potentially reasonably similar outcomes under their conditions.

Just to try to put this in context, where does that fall if the major issue of haploidentical is getting rid of the T-cells? In addition, for the cord blood comment, just to extend that a bit, the protocol to use conditioning or prolonged cyclosporine might be an institutional specific situation. There may be other groups that don't necessarily have to adhere to those particular conditions. I just wanted any comments on --

DR. BUCKLEY: Again, I think for me to do a T-depleted mud transplant, for example, whether it's cord or adult donor, I would want to do this without any pretransplant chemo and with T-depletion. But if you could really match, you know, have a 10-out-of-10 match, then that would seem to be a good choice.

DR. RAO: Dr. Murray.

DR. MURRAY: Dr. Buckley, this is enormously helpful so I wanted to thank you. I feel like we're coming to a fairly rapid and clear sense of answering the first question which is this risk benefit ratio between marrow transplant and gene therapy for X-link SCIDs.

There was one piece that didn't get pinned down that I think Dr. Blazar asked and that is for a child, say, in the French trial who develops one of these lymphoperliferative disorders, two of the children are now, I understand, in remission. If I understand correctly, one has died, but for a child with that sort of disorder, what is your sense of the likely clinical future that child faces?

DR. BUCKLEY: Well, I really have no idea. We thought that the first child who had first gone into remission and then I think he had a recurrence and then he was given a chemoablated MUD transplant. One would have thought that perhaps the chemoablation and the MUD transplant would have gotten rid of his underlying leukemogenesis. The fact that it came back is the thing to me that is very worrisome after that type of treatment. I don't really know if these other two children remain in remission, then I would think that certainly minimizes the seriousness of it.

DR. MURRAY: Could I follow up?

DR. RAO: Go ahead.

DR. MURRAY: Although the FDA has asked for our response specifically in relation to the X-SCID trials right now, we've also heard about the ADA trial where I gather the natural or the experience with transplant is not as favorable so I wonder if you would answer the similar question I think Dr. Mulligan phrased, if you had a choice between a gene therapy for ADA-SCIDs versus marrow transplant.

DR. BUCKLEY: Just based on the experience so far, I would say that gene therapy appears to be a very good form of therapy for that condition. I would -- that was the point of my saying these are 10 different diseases. I don't think you can conclude something from one to another.

DR. RAO: If I can take it one step further, most of the results with the adverse events that have been reported, and there are explanations for that, but perhaps were from one study. There is another ongoing study which has treated patients and has followed them for at least 24 months and maybe now 30 months and there hasn't been an adverse event. Would there be a better comfort level or would it change your opinion depending on what was the report from a second independent study in the ability to generalize even among the SCID population and the treatment with gene therapy?

DR. BUCKLEY: Well, I think that is very important. The British study, I think, is still open. Isn't it? They are still recruiting patients I believe.

DR. RAO: Can I have a clarification? That was my impression.

DR. WILSON: (Nods yes.)

DR. BUCKLEY: And if they continue to have no similar leukemogenesis in their trial, then that would make one want to go back and look at what it was about the type of gene therapy that they did in France as opposed to England. If that trial continues to have no problem, then that would certainly change things, I would think, in the future.

DR. RAO: Would that be the sense from the other members of the committee as well? Dr. Mulligan?

DR. MULLIGAN: Well, I was just going to turn this back to the ADA case and say that from the study we did hear about, the one from Italy, my impression is that there are two patients that were five years out okay and then the rest of them were clustered around the 30-month point.

I guess my question would be why would we think that if there's going to be some comparably catastrophic event that would occur in that class of patients that would happen at 36 months or something, there's no reason why that would be the case. If we in hind -- if we looked at the case with the SCID, well, everyone is very happy about this. There's no difficulty.

I guess the question I'm raising is knowing that something can occur completely unanticipated at about a three-year period after transplantation, what are people's comfort with thinking that the ADA trial results represent something that is completely different? It doesn't seem to me that with the weak statistics that we are faced with that we can even draw any conclusion. Clearly the reason why people draw conclusion is the biological differences between the two. I just think you should be very careful to separate the two. We can definitely have our thinking why we think there's a great difference and I personally think there is a difference between the two. The facts are incomplete.

DR. BUCKLEY: I agree with that. I would favor waiting until a similar time period has gone by both in the British trial and in the ADA trials before making any further conclusions.

DR. RAO: Go ahead, Dr. Salomon.

DR. SALOMON: Just as a point of explanation, exactly how many months is the British trial now in terms of mean follow-up? Because what is really striking about the French kids is it's between 30 and 33 months which I think someday someone will explain that and it will be really, really interesting.

DR. WILSON: In the December Lancet paper the longest time to follow up was 29 months.

DR. SALOMON: Yeah, that's what I remember. You figure it's not quite ready to interpret that study but it's a cautionary note because I sit here and you go, "Oh, okay. Here's intellectual clarity. We should definitely come up with a strong statement against X-SCID gene therapy. But think about it. Put it in the context.

If we had done that, then the English study wouldn't have gone forward. Not that we have any regulatory authority in England, of course, but you get my point. Then I don't know. Have we moved the field forward? Have we done the right thing? No, but I don't know how much clarity we have at this point. The mechanism of these things is still really confusing, really complex.

DR. RAO: Jon.

DR. ALLAN: I wanted to ask Richard a question because this thing came up about the English study versus the French study. Have you looked at the different vectors? What is the vector difference between what the British group is using and the French group and do you think that would have any -- would that give you any reason to think that there would be any difference in terms of probability of leukemias?

DR. MULLIGAN: I guess we are going to come to this, right? The simple answer is I don't think so. Nor do I think that you could -- I mean, other than principles all we're going to see is that there's principles. I think principles of how the fewer numbers of cells, fewer number of integration sites, the less manipulation of the cells, etc., etc. These are things that are principles that you think might be important but I don't think when you look at the way one group has done it versus a second group and see the different results it's actually very helpful. I think the best thing is just your own sense of what is the best way to do this and that is, I assume, what we can talk about. What is the best features of vectors to have but there is nothing I can see that would clearly pinpoint something.

DR. BUCKLEY: And the authors of the paper in the discussion also said that they didn't think that what they had done was very different from what they did in Parish.

DR. RAO: Dr. Murray.

DR. MURRAY: Since I'm the one who used the term clarity, I mean clarity in reference to the first question we are asked, namely the risk benefit ratio between the gene therapy for X-link SCIDs and transplant in Dr. Buckley's head. I think that is a very important source of clarity.

Second thing I feel that I've learned, and I'm inclined to agree completely with Dr. Buckley, that it's a mistake to treat all varieties of SCID as identical. We've got some larger number of disease, maybe as many as 10. For the FDA, I think, each disease needs to be regarded on its own terms. For me those two forms of clarity. Everything else is mysterious.

DR. RAO: So let me try and see. I think Dr. Mulligan tried to push this as well in terms of thinking about this. When the trial was initiated for X-SCID there was at that time an alternate therapy available and there was a risk reward ratio and a risk benefit ratio and a choice between therapy that was made at that time as well.

We're not here to judge whether that was good or bad. What we are here to ask is whether that choice has now changed at this given time because we have now learned of one additional patient. The issue to me was that one additional patient that you've learned about has that affected your thinking about this in terms of just that one particular study or has it affected your thinking about all X-SCID studies, or has it affected your thinking about all gene therapy studies?

My sense from listening to people was that it's very hard to generalize from one kind of disease to another so one can't say that our thinking should change from this study to all other kinds of SCID diseases. Even within this X-SCID studies we have to be careful because all the results are from one trial and we don't quite know whether this would generalize even to other studies even if they are overall very similar.

One would be very much more worried when one followed it if it turned out there were additional cases in another independent study even if it were similar and that would cause something to change. Let's take that question and ask what if there was one additional disease or adverse event in the same study. Rather than three, as Dr. Mulligan pointed out, it turns out to be four, five, or six. Would that really change given we know that it's experimental therapy and it was high risk? Would that be the same as finding one additional case in a different study?

Go ahead, Bruce.

DR. BLAZAR: I think one thing I've taken away from this is based on the data presented on ALK2 receptor gamma-c integration sites and the potential relationship to that and leukemogenesis in the context of LMO-2, another second hit. That's the data that we speculated on before but it seems that has crystallized more now.

I think that if one takes that as a finite risk, then it doesn't really matter whether another patient develops or not. The risk is present. The same thing with the different trial. I think that it's really the strength of that observation which was a speculation before. As more data is accumulating it makes me more nervous regardless of whether we see another case or not.

DR. RAO: Also, it looks like currently model systems exist to start beginning to estimate that risk as well. Go ahead, Dr. Mulligan.

DR. MULLIGAN: Another way to address this is I think, as I recall history, when we last talked about this, I think the issue of what happened if the next case surfaced or how that would change people's thinking and I thought -- certainly I felt that I fully thought there would be another case or another couple of cases and I thought there was general consensus that was true.

I'm not sure that I see that much difference when we're sitting here talking about this than the previous time except that I would agree with what Bruce just said. I think that this smoking gun of the gamma chain is much more clear-cut. Again, to go back to distance history, this was not an issue that was never talked about.

People years ago very much said that there wasn't going to be. This is just a kind of bit error in IL-2 signaling. Warren wouldn't say that but certainly I remember speaking to the French group and I think there was not at all a sense that there would be a risk factor from the actual therapeutic gene.

DR. RAO: Let me just ask one more question here. Do you feel that the third case doesn't look like LMO-2 and there doesn't seem to be clear data that it's the gamma chain either so that may be a different combination of genes that are affected. Right?

DR. MULLIGAN: Yes. I guess I would say there was one other thing that I felt was significant from the previous session and that was I think there is a greater appreciation for the number of proto-oncogenes or sensitive locations. That has been something that has occurred over the last five years or so.

There are many places that are probably not good places to land your retrovirus sequences. To me if you're asking does this make it any worse that this is not the LMO-2 gene that this latest case refers to, it doesn't seem to have any impact on my thinking.

DR. RAO: Dr. Leonard.

DR. LEONARD: Related to part of your question, I guess if it were another patient in a different study, in the British study, versus another in the French study, that would bother me more and I think would probably bother everyone somewhat more because although we wouldn't understand the basis. Right now there could be something strange or specific to the French study that no one can explain where it is conceivable that a different factor and slightly different protocol might have a much lower risk of these types of events. There also is another aspect that for me is a little different from several years ago.

This might be something that Becky might want to comment on. One of the early thoughts related to the advantage of gene therapy is that there was never much of an issue when one succeeded with bone marrow transplantation in engrafting T-cells in natural killer cells. But there is the variable nature of the extent to which one reconstitutes the B-cell compartment and the need for chronic intravenous gammaglobulin.

But my understanding of that, and you should comment, is that, in fact, there has been better progress and less of a difficulty with the nature of B-cell end of problems versus what I and perhaps others thought it was some number of years back. Why don't you comment on that end.

DR. BUCKLEY: Well, for the X-link and the JAK-3s it really hasn't changed because the B-cells that are host derived still have their six abnormal cytokine receptors that they can't signal. Unless you get some donor B-cell engraftment, you're not going to have B-cell function in X-linked or JAK-3 deficient.

However, in about a third of our X-links we do have donor B cells and so they are off of IVIG. We don't know what's different about those children than it is about the ones who didn't get that.

In the other type of SCID, for example, the CD3 gamma -- I mean, sorry, CD3 delta and the CD3 epsilon and the IL-7 receptor alpha, they have normal B-cell function with their own host B-cells so their B-cells work. I think it really depends on what the molecular defect is in terms of whether you are going to get B-cell function or not.

DR. RAO: Dr. Salomon.

DR. SALOMON: I was just going to bring us back to the point where one question here is is this vector glass driven? In other words, that has implications in our thinking with respect to other types of retroviral trials.

Again, thinking about the history of this, that was what created such a big issue the first two times we dealt with it. How far does this ripple, the stone into the pond, how far do the ripple go out to affect other retroviral trials? I think we need to think about that.

If to the extent that it's being driven the evidence is still far from solid as Bruce points out, too. The point is if it's being driven in part by the common gamma chain, then we could say perhaps it's more specific to this kind of SCID and that would significantly change how far we worry the ripple is going out from the center.

The last point I want to make is what I see yet is that when they look at the mutations in leukemias so you look backwards at disease, you still are seeing an incredible number of hits. This idea that there is one or two hits that are occurring, I'm just not sure whether we really have a very good handle on that.

If there's a whole lot of hits, then that also has implications on how you do the gene therapy and the construction of the vector and the force of the driving, the promoters being used, etc. And to what extent is the therapeutic gene and to what extent is the vector's construction driving this in a state where there's a whole bunch of other endogenous naturally occurring hits.

DR. RAO: So we should keep that in mind for the subsequent set of questions as well.

DR. ALLAN: That is sort of like where my mind is going, too, which is when I was here last year, you know, we said there's two kids and now there's three kids. We've learned more about transcription, where integration sites go.

They go into transcriptionally active sites. What I started to think about is, well, you look at HTLV or something. It actually acts in some ways like a vector system because it doesn't replicate very well and it has its own oncogene.

It takes a couple of decades, maybe even four decades before you see leukemia so you might want to -- you could argue that in cases where you don't have the gamma chain that it may take four decades to see the leukemia without that gamma chain transgene just where it's integrating. These are the kind of things I worry about.

DR. RAO: Bruce.

DR. BLAZAR: And to integrate those two comments, I think, a case in the British trial would be of even more concern but the gamma chain is integrating somewhere. We don't know how many copies it is. We don't know in all cells where that is. There's a huge proliferative advantage to the integrated cells.

Not only homeostatic proliferation but gamma-c driven proliferation. I think that one can't get beyond the concern that this is still a unique situation of regardless of the vector having a finite risk of a gamma-c integration driven process because we can't control copy number sites and there is such a tremendous proliferation. The question is whether there is any analogous concern to other cells that are undergoing selection in vivo from trimethotrexate or anything else.

I think that discussion needs to happen at some point but it's really whether selective events are teasing out cells that have proliferative advantages that may have already had a single hit where these are providing a second. More than I think it's necessarily just the vector because we don't have site specific integration processes. Richard doesn't like that.

DR. RAO: Barbara and then Richard.

MS. BALLARD: One piece we're kind of leaving out of the puzzle, though, is we're comparing everything with Dr. Buckley's studies. While she's got the longest running and the most number of SCID patients in the U.S., it's not the only program unless every other program does use chemotherapy. Chemotherapy in and of itself can cause cancer so we are not comparing those other studies and their incidence of cancer and leukemias to this. We are leaving that piece out.

DR. RAO: Yes. What I'm trying to say is that the risk benefit ratio has been modified best by looking at the statistics that Dr. Buckley presented so you are absolutely right. It was just -- that's why I tried to point out in the beginning that at some point the decision was made which was rational and reasonable at that time that gene therapy is important in this class of patients. We don't want to change that and say that was wrong. All we were trying to say was that in the light of new information does the risk benefit ratio change.

I think the question was has their been significant improvement in alternate therapies at this time which would merit a reevaluation. As Dr. Buckley pointed out, it hasn't significantly improved. There hasn't been a major breakthrough on that side which would suggest that but that they have good results. We should keep that in mind.

DR. BUCKLEY: Well, if I could just say the one caveat, though, is if you can have an early diagnosis. We have a greater than 95 percent survival if they are transplanted in the first three and a half months and only five of those were HLA identical. The rest were all haploidentical. I think that is the new information that if you can have an early diagnosis and transplant early, then you have a much better outcome.

DR. RAO: Given the current requirements that one can consider gene therapy in children below a certain age given what happened, it might be useful to keep that -- it wouldn't be relevant for that specific issue.

Dr. Mulligan.

DR. MULLIGAN: I think the issue whether gamma chain is different than other things, in fact, will be best -- maybe actually best addressed by doing these other things. I think we have seen the information that we got that we didn't have from all the mouse work that we did for years. We really should be almost encouraging, I think, other bone marrow retroviral transplants to see whether or not that will tease out whether or not there are the same things happening.

You could imagine it both ways. You could imagine that you will convince yourself and salify the fact that gamma chain is very, very different, or you will begin to tease out, I think, the things that Bruce was pointing out, you know, proliferative selections, those kinds of characteristics of diseases that make them worse candidates.

DR. RAO: Go ahead, Bruce.

DR. BLAZAR: What I was advocating for is to continue the other trials and the question is what for the gamma-c trial or the proposed trials here, what is the risk benefit ratio there. I was trying to make the point that seems to be in a different class. It's a class of cells with a tremendous proliferative advantage potentially.

It's at least a testable question but I think that the gamma-c from what we've learned from the mouse studies now plus the human experience does put that, at least with current knowledge, at a much higher risk apparently than what we have with current knowledge for other studies. I was only focusing my comments on whether gamma-c should go forward at this point. I feel that would be a pretty high risk.

DR. MULLIGAN: I was trying to make the point that I don't think we have data to compare it to. There isn't really much of any other data other than this bad outcome that we see because the other trials are not as mature as this trial or they are just becoming -- getting to that point.

DR. BLAZAR: I think we're saying the same thing.

DR. RAO: Dr. Harlan.

DR. HARLAN: I thought earlier -- I think it was you, Bruce, that raised the point that anybody with this condition, X-link SCID, that can't or has failed a bone marrow transplant is a perfectly logical group to continue these types of studies because they've got no place else to go.

DR. BLAZAR: I was referring to up front utilization and then I was trying to push the point with Dr. Buckley with whether a failed haplo transplant would necessarily exclude trying to do a matched unrelated donor to depleted or cord blood. I think that is the gray area that we never really reconciled as a group.

DR. RAO: So I think we've really discussed this question in quite a lot of detail. To me it seems like one major consensus is that this doesn't change the sense of unease dramatically. I mean, it raises the sense of unease with what is happening.

It says that one should monitor and follow this closely and, in particular, follow the second more related trial which is going on in England as closely as possible. But there are two things that have happened which allow us to feel some sort of sense that there is additional information from the animal studies that have been done in this sort of really detailed mouse models that have been analyzed which suggest that there may be a basis which might suggest that it may be a little bit more specific in X-SCID in terms of the mechanism.

There is not sufficient data to be able to make a strong conclusion. It wouldn't change our sense of unease dramatically if one more patient died specifically because that is an extension of the adverse event rather than a new adverse event.

However, I think there would be a larger sense of unease if there was a new case in an independent or different trial. Does that seem to summarize the sense of the committee? It's clear that there is a need for additional information which would allow one to get a better assessment of the risk.

It might be better than what we thought it was given that we know now in that it's more than one hit or a multiple hit and there might be ways to circumvent that which we may learn about when we think about vectors. The additional data hasn't suggested that there is a heightened risk. Does that seem like a fair assessment of the data?

DR. SALOMON: Yes. And I think the other thing that has come out in the conversation here is we talked about there is a hit from the retroviral vector that is vector class specific. Then there are endogenous hits and that may or may not be influenced by the disease, the age of the child, the type of disease.

Then another thing that comes out is that the therapeutic gene, in this case the gamma gene, may be an important factor. In a sense there is even a finer distinction there because some of the therapeutic genes, like the gamma chain, are growth factor receptors much in the same way that the truncated nerve growth factor receptor transduced a signal in partnership with another alpha chain.

Some of these diseases will not be, however, like ADA it's not a growth factor, it's a survival factor, and we don't know what the impact of that is but it's likely to be very different than a proliferative transforming growth factor impact.

It is a really interesting thing. My feeling also in consensus with what we've discussed is that we should be very cautious not to put any unreasonable obstacles in the way of going forward in some of these other areas. I think that is really important.

DR. RAO: Carolyn, did you have something to add?

DR. LEONARD: So I was going to say that the cytokines that use the gamma chain are both growth factors and survival factors like IL-7 has major actions as a survival factor. It really has the possibility of doing two things. Understanding how it's involved is something that some of us have spoken about.

Understanding how it might be involved is what I meant to say. One thing is that in the X-SCID there is an empty T-cell compartment essentially. Once the gamma chain is provided the possibility of a very large homeostatic proliferation to fill the compartment.

It might be that is the type of setting where something like the LMO-2 oncogene might be able to theoretically have a greater effect. Although there are certainly very different -- SCID is a spectrum of many different diseases and I totally agree with Becky and others that we have to think of them separately.

JAK-3 deficiency is, in fact, very similar in the clinical and the immunological phenotype is indistinguishable except it's in both boys and girls. JAK-3, of course, physically associates with the gamma chain and mediates the actions of each of the six cytokines in part until one of the questions perhaps to come back to is how we feel about JAK-3 deficiency and is that to me thought of a little more closely to X-SCID or not and I can see it both ways.

One aspect of the gamma chain is that it's constitutive, or so we think, but the idea that constitutive expression of the gamma chain might be important is certainly a possibility because our knowledge of the gamma chain is, in fact, primitive and we don't really know if it's down modulated and even absent in select populations at different points in times.

In particular, in the thymus and so on where maybe we haven't been able to look in the right way collectively as a scientific community. Maybe JAK-3 is not that way in the same way. It is regulated. There are many distinctions.

Anyway, I just throw that out that the clinical spectrum in disease is very similar, yet there are differences in what we could think of as biological regulation of the gamma chain in JAK-3. I don't know what the right answer is but it's something that I think should be thought about.

DR. RAO: And I think you summarized that very well for us and that would probably be exactly what the committee would say even if there was additional discussion on that topic.

Let's go to the second question.

DR. WILSON: I just wanted to ask for one clarification on the consensus that you put forward. When you said that the committee doesn't have an increased rate of unease, as you put it, with the third child, I just want to make sure I understand that, therefore, you would recommend that the X-SCID gene therapy clinical protocols continue without modifications.

Or going back to some of the earlier points that were raised where I thought people were recommending only in certain circumstances, for example, failed bone marrow transplants would you consider it? Could I just get clarification on that?

DR. RAO: Go ahead.

DR. SALOMON: I thought the answer to that would be that right now in X-SCID gamma chain gene therapy trial should stay on hold. I thought we were clear about that. If I'm not, if that's not true --

DR. HIGH: Even in the setting of somebody who failed bone marrow transplant?

DR. SALOMON: I think that would be a different situation. I think that was the point that was made. Thank you.

DR. WILSON: Perhaps if you wouldn't mind, Dr. Rao, trying to craft a consensus on this one point so it's clear for us, and make sure that the committee agrees with it, if there are any circumstances whereby gene therapy for X-SCID would be acceptable.

DR. RAO: Do you want to, Kathy, propose a compromise?

DR. HIGH: I would like to ask two questions first. In the 132 SCID patients that Dr. Buckley wrote about in the review and there were 30 deaths, were they mostly early on or were they scattered evenly throughout the years of follow-up?

DR. BUCKLEY: They were mostly in the first year and we had, I think, six late deaths mainly from persistent CMV or EBV. They were mostly from infections that they came in with at the time of referral.

DR. HIGH: And then the other question I had before I propose a consensus was what percent of cases are diagnosed in the first three and a half months of life because you said those do so well.

DR. BUCKLEY: Well, we have done 39 and we've transplanted 141 so you can do the math.

DR. HIGH: So not that many?

DR. BUCKLEY: Well, we hope it's changing because right now one of the things that I'm lobbying for is newborn screening for SCID because if you could diagnose this at birth, then you would have a much higher chance of saving these children.

DR. BLAZAR: Before you go to your consensus, why does the cumulative experience with failed transplants with regrafting with other sources just worldwide what is the experience with MUDs, peripheral blood, bone marrow, or other sources?

DR. BUCKLEY: I really don't know those numbers, Bruce. I can just tell you that out of our 141 we had done those three cord blood transplants and two of those died.

DR. BLAZAR: After failure worldwide so if you fail the first haploidentical, what is the outcome data?

DR. BUCKLEY: I don't know what the overall data are on that.

DR. BLAZAR: Because that's going to be the discussion for the option is depending on those data.

DR. RAO: Barbara.

MS. BALLARD: I was just going to respond to Kathy. The problem with diagnosing SCID currently is that until the baby is almost critically or fatally ill it's not diagnosed unless there's a family history. In an X-SCID you do tend to have family history by being an X-linked disease but even then it's not always diagnosed before the baby is critically ill.

DR. RAO: Go ahead.

DR. HIGH: I recall Dr. Cavazzana-Calvo when she was here two years ago talking about more rapid reconstitution of T-cell function in the children who had gene therapy as opposed to the normal kinetics for bone marrow transplant. I believe that I heard Dr. Buckley say that there wasn't that much difference.

DR. BUCKLEY: If you go back to the original paper they published in Science, they have a figure in there and it shows it takes 90 days before the T-cell function comes in.

DR. SALOMON: Kathy, I think that what you're thinking of is the dramatics two years ago was that it was the B-cell compartment that was reconstituted more efficiently in the children in the gene therapy trial and that was, to me, a really important point at the time because we had people come and testify about chronic administration of IDIG and the fact that even though the statistics look good like survival statistics that the reality was darker than it sounded because these kids were on chronic IVIG and having low-grade infections, etc. I think it was the B-cell compartment that was dramatic and NK.

DR. RAO: Go ahead.

DR. KOHN: On the issue of reconstitution I think in the French study it was more like 60 to 90 days. When we do haplos it's more like 90 to 120 days before we see T-cells. Ken Weinberg's X-SCID study is now modified to only include kids who come in with bad infections like CMV or edno because of this quicker immune reconstitution. I think that's what you are remembering.

DR. BUCKLEY: That figure from the article but I do have it -- you know, I could show you a PowerPoint view of it. Do you have the one showing the proliferation? Okay. Right. No, it's the proliferation slide. It takes 90 to 120 days before you see function come in.

DR. RAO: Let Kathy see what you've seen.

DR. HIGH: Well, this is a critical point of what I was going to propose because it seems to me that I always favor there being as much choice for patients as possible, as many options for patients as possible, and present to them what the options are. Certainly it seems to me that if they failed bone marrow transplant that would be a setting where this procedure should be available. Then the other issue is if the baby does come in infected and is likely not to do well at bone marrow transplant, would that be another case where that would be an acceptable entry criteria for a trial.

DR. RAO: I think we should distinguish between criteria for a child and for what would happen in one of these current trials which are on hold, I guess. Right? Is that part of the question?

DR. HIGH: Would that be another instance where this should be made available.

DR. SIMEK: I think they are not actually separate at this point because some of these trials take into consideration failure to bone marrow transplant so we actually would appreciate a consensus and have some exceptions if need be because it will affect these trials.

DR. RAO: Go ahead, Dr. Salomon.

DR. SALOMON: The reason that I didn't bring it up initially was, and this is something I'm looking to input then from my colleagues, particularly pediatric bone marrow transplanters.

My thinking would be that once a kid goes through a whole series of bone marrow transplants on the way to being designated as a failure, the bone marrow transplants, from an X-SCID, that very few of those kids are going to be in any kind of condition to now undergo an autologous bone marrow transplant with gene therapy.

I wonder whether we are putting like a lot of emphasis on something that is not likely to be an issue maybe once in the next five or 10 years. I don't know whether this is even worth the time. I could be wrong.

DR. BUCKLEY: If I could make a comment. The summary that I had to be copied, we had 10 patients who did not have T-cell chimerism out of the group that was presented. Of those 10 patients seven of those were ADA deficient. One of them was an X-link SCID and -- let me see if I've got it right now. One was a RAG-1 SCID so there were only -- most of the ones who were not chimeric with the donor cells were really ADA deficient. It would be pretty unusual to have an X-link SCID not engrafted from a haploidentical or HLA-identical transplant.

DR. RAO: So would it then maybe be fairer to say that in cases where no other alternative exist in terms of giving patients a choice?

DR. BLAZAR: I think you have to define alternative. You mean reasonable alternative.

DR. RAO: Reasonable alternative.

DR. BLAZAR: I think the question is for the purpose of entry criteria even if it's a rare event. If someone wants to have an open protocol to have a therapy option and they failed a bone marrow transplant and they have X-SCID, should they be offered gene therapy?

Even if it's a rare event, if there is currently a protocol that says high risk for a haploidentical transplant and another one that says previous VMT with persistent T-cell and B-cell impairments, the question is what to do with those protocols.

Specifically, and I think that if they failed and they meet other entry criteria for however that's defined to have enough health to go through the procedure that this is a therapeutic option. That's what I would make as a proposal for discussion.

DR. RAO: Dr. Salomon, what do you think? Dr. Mulligan? My sense is that should be the case.

DR. WILSON: Thank you.

DR. RAO: Question 2. "Please comment on what changes, if any, would reduce the risk to subjects in clinical trials using retroviral vector mediated gene transfer in X-SCID. Please consider the following: Limit the dose based on total vector copy number in the transduced cells. Limit the dose based on total number of transduced cells and alteration of retroviral vector design."

I guess here it's fairly important to take into account the new evidence we heard in terms of what happens with integration and the frequency and the theoretical calculations that Carolyn pointed out as well. I would also say that my sense was that it's still uncertain and, as you pointed out already, it's hard to say anything other than general principles.

One can't give an absolute number which would change what was already suggested or what is already introduced in the trials. Right? If you remember from last time, there was a reduction in the total viral load that you could use and the total number of cells that would be put in was also changed in the French trial. Maybe one could start off with some general principles and I can ask Dr. Mulligan maybe to make a statement on those lines.

DR. MULLIGAN: Yes. I think we should really strongly encourage people so don't get me wrong. I just don't think that there's any legislation possible. There's no real things you have to insist upon. I would say of all the different approaches, the one that I personally think most doable is suicide systems.

I think the insulator parts of things, the way people test them are usually wishy washy. They are very difficult to test properly. You might argue it's better to have them than not to have them. They don't probably do any harm but I think it will be a long time until the FDA gets convinced that these things really will have a major impact.

We don't even know, I don't think, in the case of the insertions here whether it's actually an enhancer of fact or whether it's a disruption of sequence. I think in the case of the suicide systems which, I think, are not ready for prime time at this point, there are experimental tests that ought to convince people.

There are ways to test things using tumor implants where you can really show that you can kill this number of tumor cells and so forth. I think ultimately the ability to abort the mission is going to be very important. In fact, I hesitate whether some exception to the X-link SCID advice we gave would be that if you did come up with something where you really could kill the cells, that might help you.

I would probably not go quite that far at this point but say that everyone knows the principles. We can go through them but, I mean, the idea of limiting the total number of integrations is obviously going to be important. I still like to use the old quote from many years ago about if no gene transfer occurs, gene transfer is perfectly safe.

I think the difficulty is that we're going to see this. If people begin going easy on a virus for the infection, probably you're not going to see good things happen. The more gene transfer that occurs, the better off things are going to be. Also maybe the greater risk you're going to have. I think just knowing what I know experimentally to try to figure out how to limit the number of insertions is going to be not a good idea.

I think if you wanted to address that issue, you might encourage people to actually do a dose estimation curve where you go from -- this was never done with the number of patients and so forth but the idea would be to lay off the virus and then go up as high as possible. I mean, some sort of experimental approach like that as opposed to, I think, now people are just trying to hit things as hard as possible.

DR. RAO: Is there a measure that one could use to look at number of integrants or that technology doesn't exist because then maybe one

could --

DR. MULLIGAN: You could presumably do infections for transplantation and get some sense of copy number. I think it's too complicated. I think it's not going to be a practical way to proceed. So in terms of limiting the number of cells, I mean, to me that's almost a bone marrow transplant question and that will definitely be a characteristic of the disease.

You know, how many cells you have to have for reconstitution. I guess I would finish and just say strongly encourage these things. I don't think we want to legislate any real direction. There's a lot of creativity in terms of vector. People are trying different kinds of things. They think we ought to just make the point that here are a bunch of different issues that we think are very important and will make for a safer overall approach.

DR. RAO: Dr. Wolff.

DR. WOLFF: I think we all agree that probably getting new vectors designed would be the best thing of all but that's a long ways off. Right now I guess we could deal somewhat with the number of copies. At least maybe it's hard to quanitate exactly but it is kind of scary to think if we keep escalating the multiplicity of infection to keep trying to improve the efficiency.

I think that's not a good idea to overdo that and put some limits on the multiplicity of infection. Based on Christopher Baum's work, I think clearly we don't want, I would say, more than two integrations per cell. If we can somehow limit that it would be a good idea.

DR. RAO: As far as I could see, and maybe experts who present it can also tell us, that there is no simple way to measure your frequency of integrants and say that is true and whether there will be selection of a particular clone, which was rare, otherwise which would do it and which of those clones would really integrate it.

DR. WOLFF: In Dr. Baum's work, and I think you should talk about it probably more than me, but my understanding was that he used either just supernatant versus co-cultivation and that made a big difference in the multiplicity of infection or the number of integrants.

DR. RAO: Dr. Salomon.

DR. SALOMON: I was just going to bring up again something that we felt strongly about before and I still feel strongly about is that it would be unreasonable. In fact, you would really put a terrible hit on gene therapy going forward if we started insisting on a lot release criteria that everybody measured the number of integrants or where you integrated. We discussed that the last time, too, and felt at the time that was just as unreasonable two years ago as it is today. That's still my opinion unless somebody around here --

DR. RAO: Is it fair to say that there has been no new magic technology or no new strategy which would allow us to put that kind of limit or be able to try to put that kind of limit?

DR. MULLIGAN: I would even say even if there was you wouldn't want to particularly. I just disagree. I don't think that's the right way to approach it.

DR. SALOMON: One of the interesting things that came up at the time that I suggested, if somebody remembers, the suicide gene idea that Richard brought up a moment ago and there are a number of different ways to consider doing that. At the time the interesting thing was the FDA leadership was aghast at the idea of any sort because all of a sudden now you have bicistronic vectors and that would significantly change the whole situation.

DR. RAO: Here is another sort of question for the committee as well. If you can't really look at multiplicity of infection any reasonable way or you can't really look at the number of integration sites, is it even worthwhile to consider saying, well, you know, from these lots there are screens. We know there are 150 candidate oncogenes.

We should really look to see if there's a candidate integration. We just don't know enough to be able to rule that out because, as you pointed out earlier, that might be a positive in terms of helping the cell without it being a transforming event. Is it worth for the committee to make some statement saying that it's not possible or not reasonable at this time to look at those?

DR. MULLIGAN: Maybe I'm missing something. I mean, you can simply get the gross total number of copies. That's easy. Right? I mean, if you wanted to legislate which, I think, would not be a good idea, you could say you can't have an average copy number of over such and such. Right? But there's obviously a poisson distribution of things. You can do that very simply.

DR. RAO: That doesn't seem to be very useful.

DR. MULLIGAN: I think the problem with this is it's obvious that the more copies the more risk. That's clear. It's just whether there's some more sophisticated biological effect having to do with combinatorial mechanisms and so forth. The simple version is just the more copies you have the more at risk you are. I question whether, first of all, practically you really should limit. Second, I think, for development of a theory you need to know how much therapeutic effect is going to be helpful.

DR. RAO: Dr. Buckley.

DR. BUCKLEY: I was just going to comment couldn't those studies be done in the knockout mouse or in the dog to determine how many copies and how many cells you need to correct the defect?

DR. MULLIGAN: Yeah, you can. I mean, I think you can do that but I'm not sure it's an easy to translate to the patient part.

DR. BUCKLEY: No, it isn't easy to translate but I think we've already seen from the limited number of people who've been treated that if you give a very low number it doesn't seem to work very well. But you really don't know what the optimal number is and if you really wanted to do a titration you would have to do it in an animal model.

DR. MULLIGAN: I think it's a much cruder art than you might think, even a mouse system where if you look at the collective work of people, there's certain people that are very good at this and can do a viral infection transplant and get very good copy number and so forth.

Then there's a world of people that are having difficulties getting the high copy numbers that you may not want to have. I think that, in fact, in the clinical trials we are going to tend to be for still some time probably at the low end of this. We'll see but I think that we're not going to actually get up that high.

DR. RAO: Dr. Kohn.

DR. KOHN: Just a technical comment on trying to use copy number in the transduced cells as a lot release criteria. If you use realtime PCR to measure that, you often over estimate because there's a lot of unintegrated forms, LTR circles that won't go anywhere.

In our clinical trials when we look at that we typically see the transduced cells have an average of three copies per cell on average in the a number of studies. But if we then grow colonies, it's about 30 percent so it's 10-fold higher than you're getting by colony so I don't think technically it's easy to look at the product as a test release criteria.

DR. RAO: Go ahead, Dr. Baum.

DR. BAUM: We should also consider that with current technology there is a biological limit so that stem cells don't accumulate so many copies when they are transduced. That has actually been shown by Hans-Peter Kiem's lab in the nonhuman primate model. With current technology I wouldn't be afraid of having multiple copies in stem cells. But with future technology there might be an issue but we have enough time to test that until this is moved to clinics.

DR. RAO: Go ahead.

DR. BORDIGNON: I think that the most sensible recommendation lays in the work that Dr. Mulligan used before. It's very difficult to regulate this in a very tight fashion because, of course, let's say you want to reach a given level of a protein but you cannot fit more than two integrations. Let's assume that you use any of the parameters. Then you may find yourself or the investigator push toward increasing gene expression so utilizing stronger promoters.

In a way I think the only way that all this reconciles is that you can give recommendation so that specific communities will evaluate the composition of all these elements. These elements are relevant for the risk but then I will let the investigator to design the best combination, the safest combination and then the committee evaluate whether this was analyzed properly.

On the side of utilizing this type of animal models for specific correlation, we should always keep in mind that when you move from human to mouse, the type of vector you have to use changes. It changes the envelope, changes the characteristic and so on. It's very difficult then to extrapolate the data from one case to the other.

DR. RAO: Go ahead, Dr. High.

DR. HIGH: I want to return to one point that Dr. Mulligan raised with respect to 2(c) that has an effect on the answer to 1, and that is that in my mind I would consider a redesigned vector with any of the features that were discussed either a suicide gene or with insulating elements should be evaluated as a new case with respect to --

DR. RAO: Different risks, you mean.

DR. HIGH: Yes, different risks. Absolutely.

DR. MULLIGAN: I would be comfortable with that. I was just saying of the different things I imagine that the suicide is the most likely thing that someone could look at and say this makes a difference. This would probably improve the safety features.

DR. SALOMON: Yeah, or maybe it's not even so hard. Maybe you put a hist tag -- I'm making it up -- on the end of the gamma-gamma chain and then you've got an antibody you can nail just like NICD 20 is being used now for B-cell lymphoma.

DR. RAO: So does that seem to be -- I just want to make sure that we have consensus that it's very hard to put an absolute number or make an absolute recommendation whether it's in the MOI or whether it's in the vector design or whether it's in the number of cells or the number of transducer infected cells.

However, there are really important general principles which the committee seems to all agree it's important to keep in mind and that those need to be considered when a specific study is going to be evaluated. Does that seem to be the consensus of the committee and of the experts who made the points that they did? That was easy.

Actually, the next two questions are a little bit easier because we have discussed some of those issues already. I'm going to go through three and four. I'm going to suggest that we do four before we do three just because we've just looked at vector design a little bit.

"Given the increased efficiency of lentiviral vectors that transduce cells often resulting in multiple vector copies per cell. Up to 10 have been reported. Please discuss whether restrictions on vector copy number per cell are warranted for the use of lentiviral vectors." We just did this for other type of vectors in ex vivo transduction clinical protocols. "And, if so, what limit would you advise?"

I think the answer is similar to what we just said and I would like the committee to say whether they think that is true or whether they disagree with that statement.

Dr. Mulligan.

DR. MULLIGAN: Yeah, that's true.

DR. RAO: Dr. Salomon. Okay. Can I ask the experts who have been doing some of this work, Dr. Kohn, Dr. Dave, Dr. Wolff? Do any of you think otherwise?

DR. KOHN: I think it is going to be a relevant fact that will need to be looked at on a case-by-case basis because with lentiviral vectors you don't have this limitation so I think it will be something that should be considered as each protocol comes in and criteria developed. I think it's difficult to do it in advance now.

DR. RAO: And given the lentiviruses will also infect stem cells with a higher efficiency than they do with other sets. Go ahead, Dr. Wolff.

DR. WOLFF: I just think these should be tested a lot beforehand in other kinds of systems and models to see about induction of leukemia specifically.

DR. RAO: Is there also an issue with lentiviruses because there is some data on site specific integration of lentiviruses as well in addition to adenoviruses so is that a specific concern for lentiviruses or there is no specific concern for lentiviruses which is over and above that?

DR. SALOMON: I think the other thing that is important to frame this with is that what causes leukemia is not at this point proven to be the number of insertions in the genome so the argument that one would make that if you had two you would be safer than if you had 10 is ignoring the five other factors from the transgene to the vector class to the promoter to the construction of the LTR to the way you transduce to the disease you put it in.

I'm not trying to be absurd here but I can go on. I think that if you seize on one of these things because it sounds like a metric and then you start insisting that people meet that metric and it makes us feel good in a regulatory way, I don't think that's doing anybody a favor nor is consistent with the science.

DR. RAO: Don't you think that there is actually some data which would suggest that integration into an active zone can either interrupt or alter the expression of endogenous gene so that is one aspect of the potential problem of viral vectors of any kind. If you have more sites of integration or more chances than your frequency given the number of cells that you're hitting, that is certainly an issue. I agree all the other issues are equally important.

DR. SALOMON: If it was up to my wife, my kids would be driving Hummers because the risk of them getting injured in a car accident driving back and forth in California would be significantly less, but neither of my kids are driving Hummers.

The point is that the difference between 2 and 10 integrations in the complexity of everything else going on in leukemogenesis is not, in my opinion, understood so it's just a convenient metric but not one that I think ought to get elevated too high just because it seems like something we can measure.

DR. RAO: Dr. Wolff.

DR. WOLFF: I think animal models have suggested that you need at least -- well, most of the leukemias have 10 integrations at least, a majority of animal models. I mean, there are some rarer models where there might be a couple integrations but they are rare. I would say the majority of the mouse models there's at least 10 integrations which would suggest that you probably need to have that number of hits. Not necessarily every one of those is involved in leukemia but you need a lot of hits.

DR. RAO: So is it fair to say that, as you pointed out, that every new vector design has it own separate set of risks and for lentiviruses one risk would be multiplicity of infection as well and until we have clear-cut animal data we can't put an upper limit on the number in terms of the multiplicity of infection. That gives us some more time for question 3 then.

Does anyone have any specific comment on question 4 or can we go back to question 3 now? "Please discuss the impact of any of the -- go ahead.

DR. LU: About this lentivirus-based vector copy numbers. I think everyone hasn't talked about the difference between LMO gene-based vector and HIV LTR. Really the adverse event is the enhanced regions. It is clear to me that the enhanced region in LTR of the MLV. This totally acts on the cell of the gene.

You can see in every case of animals and in the SCID kids and some integrated in the different position opposite with directions with respect to the animal genes and it's the enhancer. Now we know that HIV is not known for this kind of enhancers containing LTR so I just wanted you guys to remember this point when considering the limit of the copy number. Thank you.

DR. RAO: So let's go to question 3. "Please discuss the impact, if any, of the assays in X-SCID combined with the development of myeloid sarcoma in the single monkey administered hematopoietic stem cells after ex vivo transduction with the retroviral vector on the use of retroviral vectors in other clinical indications and please comment specifically on the risk benefit considerations in ADA-SCID relative to X-SCID and in other clinical indications."

We did start off with saying something on ADA-SCID. Maybe I can ask Bruce to resummarize that the risk benefit is, if anything, in favor compared to X-SCID. Go ahead.

DR. BLAZAR: I think as far as consensus we discussed the outcome of ADA-SCID and the fact that we don't know what the particular risk is in ADA-SCID and there are theoretical reasons to think it may be different so that there didn't seem to be any reason to put on hold gene therapy for ADA-SCID at this juncture.

DR. RAO: Let me go around the committee just to make sure that we have consensus and if anybody wants to specifically add a comment to that, then we should do that.

Dr. Leonard, what do you say?

DR. LEONARD: I concur.

DR. RAO: Dr. Wolff.

DR. WOLFF: (Off microphone.)

MS. DAPOLITO: Please use your microphones so we can get it on record.

DR. MULLIGAN: I concur.

DR. SALOMON: Yes, I concur.

DR. RAO: Go ahead, Alison.

MS. LAWTON: I agree.

MS. TERRY: I agree.I

DR. WOLFF: I agree.

DR. RAO: I agree, too.

DR. ALLAN: I agree with the caveat. I'm not saying it's safe. He said I could add something.

DR. HIGH: I agree.

DR. HARLAN: I'm going to include the caveat that Jonathan just did, too, because the experiments that Cindy showed concerned me but not enough to say we should stop.

DR. BUCKLEY: And I believe before when we were discussing this Dr. Mulligan said we only have a few more months to wait and see whether the ADA-treated SCIDs in Italy will be affected when they reach 33 months. With that one caveat I would agree. I think that I would want to wait and see whether they have the same risk as in the X-link SCID.

DR. RAO: Please go ahead.

DR. BORDIGNON: The only point I wanted to emphasize is that the month you see only related to the last clinical trial so who is the one with the most efficacious data, with the full immune reconstitution and everything else. We treated patients starting in 1992 and we have patients circulating from 1995. The duration of the experiment, although the integration is not as high and as many as in the cured patients the follow-up is much longer.

DR. RAO: Go ahead, Bruce.

DR. BLAZAR: Yes. To take a somewhat different opinion, since we don't know if 30 to 33 months is a risk period for ADA-SCID, I feel like we should go forward without necessarily having the wait period. But if there is an event, then we need to convene and rediscuss.

DR. RAO: I think that summarizes it well for us. We can go to the second half of that question. I think that makes it pretty easy given ADA and we considered X-SCID and we considered JAK-3 related which is very similar to X-SCID and then we looked at ADA-SCID. Is there a concern in other clinical indications? Here I think what Allan said may be appropriate. It doesn't mean it's safe but it doesn't mean that the concern has changed in the case of other clinical indications.

Go ahead, Dr. Mulligan.

DR. MULLIGAN: I guess I would say this is probably the most complicated question because if you think of things like sickle cell or beta thalassemia, I mean, as you go through the -- if we're asking people to go through the risk benefit ratio for these other diseases, then you have to, in fact, synthesize and abstract everything that we've heard here and that's obviously we've decided very difficult. I guess my take on it is that there's not a threshold of sufficient information to say that it's not going to be safe and, therefore, it should be attempted.

DR. RAO: Can't you say that we're just looking at one side of the equation since we don't know anything about the benefit in those other indications and we haven't heard a summary? We can say that we haven't heard anything on the risk side of gene therapy which would cause us to think that the risk has increased significantly in these other clinical indications.

DR. MULLIGAN: Well, I'm not sure I would go quite that far. I think people thought that integration was going to have no difficulty. That is a complete different point. I think it's more complicated. I would say probably the most important thing is to educate the clinician who is contemplating the gene therapy who is not totally in tune like these guys are with what we're talking about that the landscape is completely different than it used to be.

There's clearly something that you don't want to have happen that can happen. Again, I don't know that you can stop these other things from happening but I don't feel comfortable suggesting that, oh, it really doesn't seem like there's that particularly great a risk. I think there's a very significant risk here.

DR. RAO: Bruce.

DR. BLAZAR: I guess from a concrete point of view there is a consent form that stipulates that there are risks and elaborates on what those risks have been. At least that have been seen as complications so far. The concrete question is whether there need to be consent form changes or not and whether there should be any additional scrutiny of what trials are ongoing.

I think the latter is premature to try to extrapolate from the data we have now. I would say we should at least with the consent forms that clearly identify the theoretical risks. I'm not sure what else can be done other than to point out what has been seen and maybe modify the consent forms if they need any modification.

DR. RAO: Dr. Mulligan.

DR. MULLIGAN: I think that's a great point because I remember when we were addressing changing the consent forms. People were waffling around, you know, how much to say that gene transfer caused cancer. I remember this several years ago. I'm sure the same thing will happen in the future. I think this is the key. Maybe that's the best you can do.

DR. RAO: The FDA has done that already. Hasn't it?

DR. MULLIGAN: Yeah, but have a very strong statement about the increased perception in general terms about the risk of a retrovirus insertion.

DR. RAO: Sharon, do you have a comment on that?

MS. TERRY: Yes. I certainly would want to see the consent form be as explicit as possible. I was just sitting here sort of flashing back to conversations I've had with Paul Gelsinger after Jesse's death and his concern that he didn't know about animal studies. I'm not sure you can draw an exact straight line here. There should be at least some indication that there is more information for parents if they want it before they make decisions.

DR. RAO: Go ahead.

DR. ALLAN: It's a little difficult, though, because, I mean, it's hard for us to even discuss this and what the risks are and to expect a clinician with a consent form to explain, "Well, in this study there were leukemias but in our study we don't know what the risk is and maybe the risk isn't as bad as that one because it doesn't have that transgene in it." I mean, that's tough. I don't know how you do that really.

DR. RAO: Dr. Harlan.

DR. HARLAN: Go ahead, Sharon.

MS. TERRY: That's why -- this is not directly related to what FDA would say or do but we highly recommend from the Genetic Alliance point of view of 600 advocacy groups that the communities around those diseases be involved.

It's much harder with something like we talked about yesterday because there's not really a community around joint disease but there is around SCID. That community and the support group can point parents and clinicians to experts who could help explain it. No one is ever going to understand it. Everybody here keeps saying, "I don't understand." But certainly to weigh the risk and benefit in an extreme circumstance like that is helped by a community of support.

DR. HARLAN: So I'm hearing a lot of discussion about the risk of the therapy but what we also have to weigh is the risk of the illness that is being treated. For SCID, I've made this point several times, often times there's no place else to go. But for many of these other diseases there are treatments. I think that it is potentially very risky for us to hide behind a informed consent form because patients trust the white coat so much and we can't just use that as an excuse.

DR. RAO: Bruce.

DR. BLAZAR: I don't think anyone necessarily wants to hide behind an informed consent but it's important to be very clear on the informed consent. I guess the question is whether there is enough information to try to regulate use of gene therapy under other indications. How does one define the risk benefit ratio given that we really don't know the risk in gene therapy for anything other than gamma-c.

Arguably we don't know it there either. For the other diseases I think in general people aren't moving to gene therapy on easily curable alternative therapy diseases to begin with. That may change in the future but hopefully at that time we'll have a better understanding of the true risk of gene therapy.

MS. BALLARD: When it comes to the disclosure, I mean, things need to be in reasonable language but the parents need to understand events have occurred. They need to understand that it's not a given this will work and it's not without some risk. It's as much as you can give a parent because they've got a kid that they are looking at who has a real high risk of dying so you already lost the roll of the dice right there, but you at least need to understand events have occurred that aren't what you want.

Then the parent at that point you've got to assume you've explained it in as good a language as you can. They've got to make their choices but you've got to give them as much simple reasonable language and as much of that information and history that you can give them.

DR. BLAZAR: So for the consent form that is available, I don't know if you've had a chance to look through that, does that meet that criteria?

MS. BALLARD: Pretty well. Again, it doesn't give as much on the animal studies and I think parents, at least the parents that really get the concept of how research is done, do also want to know about the animal studies. It's relevant.

DR. RAO: So in terms of the specific question it seems to be in my mind clear that one can't say that there's no change in perception of risk in other clinical considerations but certainly it's not enough to mandate to suggest a very specific change other than clearly highlighting that in a consent form or providing information in as explicit a way as possible. Does that seem to be a reasonable opinion?

DR. SALOMON: I think if anything from, at least, my experience over the last couple years with this whole thing has been that there are a couple key principles. The first is that the landscape has changed. I mean, however you get it out, you have to acknowledge that the landscape has changed here.

We don't know to what extent it is vector class specific. We don't know is it all integrating. Any integrating vector will do or only certain kinds? Is it the design of the vector? We've been over these once.

The second thing is in terms of informed consent, it's never going to be adequate and so the question is when you start talking about the animal studies does that mean you want to know all the animal studies that have been done that have been positive and then how am I going to explain that in high school language which is the demands my IRB puts on my consent form.

I don't want to veer off now into a discussion of all the problems with consent but I think we have to be really cognizant of the fact that I think you shouldn't put any -- you shouldn't withhold any information or refuse to answer any kind of questions or manipulate families or patients in any way but it's not -- but the consensus is a flawed process.

The last key principle I got out of this is we shouldn't be going forward with new therapies without it being really clear what the alternatives are and that gets to something Dave mentioned as well. Everyone should know what their alternatives are before they agree to a therapy like this and it has to be a good story.

DR. RAO: And I think Sharon pointed out clearly as well it's not necessarily in the consent but just have information available through support groups or something.

MS. TERRY: Because this comes to another issue that I'm not sure where it's ever going to get pushed in the Federal Government having sit on advisory committees for many branches, is the issue that it really shouldn't be informed consent that presumes that you're going to go forward with something. It should be informed decision-making and it's a process rather than a piece of paper.

The other integrating piece here is to look at whole systems and would we rather push gene therapy at high risk or, for example, put in place newborn screening for SCID when that is so much simpler, more cost effective, and sounds like very effective treatment when caught early which makes a -- and I just looked at the new recommendation for the comprehensive newborn screening panel and SCID is not on the list.

There are some issues there and I'm not sure how much this committee can talk to other committees because that was a HRSA committee, not an FDA committee. But I sit on all the committees and keep telling them they have to talk to each other. I think recommendations from one to another certainly do hold some weight.

DR. RAO: Point well taken. On that note, unless anybody has a really specific question, I think we can ask the FDA whether we have answered questions to their satisfaction.

DR. SIMEK: I think that's terrific. Thank you and we really thank you for your excellent advice today. We will take this into heed and it's been very useful for us. Thank you very much.

DR. RAO: Dr. Harlan.

DR. HARLAN: I would like for you to -- I'm not sure what our advice to the FDA is about gene therapy and these other conditions. Could you state that?

DR. RAO: Okay. I will. Dr. High, did you have a comment too?

DR. HIGH: No, no. You go ahead.

DR. RAO: I thought what I said, and Dr. Salomon repeated that in a better way, I think, was when he said we all know that the landscape has changed and clearly a perception of the risks have changed.

However, there is no clear-cut indication on a measurable change in risk in these other clinical indications. Rather than mandating anything, it's just really important to make sure that patients in those trials are aware that the landscape has changed so that even in the consent form in the process of decision-making they become aware of that information.

DR. HIGH: Okay. Are we finished? Before we all disband, I wanted to make one other announcement and that is on April 7th and 8th here in Washington, D.C. the American Society of Gene Therapy is sponsoring a conference to examine some of the scientific vector availability, financial resource, and regulatory hurdles that have conspired to slow the continuing development of clinical gene transfer trials.

Some of the people in the room are going to be in attendance and are going to be talking at the meeting but I want to announce along with my co-chair Dr. Salomon that it's taking place and that it's open to everybody with an interest in that topic.

DR. RAO: My apologies, Dr. High, for forgetting to do that. I think it's really important and everybody should attend.

(Whereupon, at 4:07 p.m. the meeting was adjourned.)