I think that data is after it but with these attenuated hour(?) backbones they simply do not grow well and VERO is one of them. Many of the strains just would not grow well.

DR. MODLIN: Thank you yes ‑

DR. HUGHES: My colleague Dr. Coffin is actually trying to look at the literature at the moment but I am a little surprised by the statement that it is 150 nanograms of pure plasmid DNA that initiated infection with HIV tissue culture. It sometimes appeared to work apparently as being quoted. What I am really surprised about is that you can form a tumor with less oncogene DNA than you can initiate a viral infection with. That does not seem to make sense to me.

DR. KEMBLE: Dr. Peden, I will let you respond but just to be clear it is not, that is not the minimum amount that can be ‑ you can go down to I believe a picogram ‑ it is that 150 that was shown when you digested it the infectivity could be eliminated. So it is that bar that we are modeling.

DR. HUGHES: So again I will echo what my colleague said earlier that that depends on the digestion really being uniform and that you have what is truly a profound distribution of fragment. I think it is at least conceivable that you may have a small amount of DNA in some sort of precipitate or a protected state and like Dr. Coffin I would suggest that PCR or some other method be used to supplement the data. The data you have is nice but I think rigorously ruling out longer pieces would be again in your interest.

DR. MODLIN: Dr. Peden did you want to respond?

DR. PEDEN: I certainly can respond. I am going to present some more of this during my talk. It is true if you take an infectious clone of HIV with one picogram in an in vitro system. So what we did for that was to mix equal amounts of infectious DNA and jerk(?) DNA mixed and then digested it and then analyzed many fractions of that and in fact I will show you the data that gave rise to that so that is where that 150 which is now 750, that is where it came from.

DR. DEBOLD: I know this question is going to come out from my constituency but can you explain why the use of the human cell lines was not preferred?

DR. KEMBLE: Very simply the two most obvious human cell lines the MRC‑5 and WI‑38 simply would not support ‑ we could not grow our vaccine strains in those cells and that is consistent with what we know of human fibroblasts. They do not go through very well. As far as using other human cell lines we actually did not consider that because in many ways the risk of adventitious agents is more problematic with a primary human line than it is with even for example our MDCKs.

DR. MODLIN: Why don't we go on with Dr. Peden's presentation? That will certainly be the appropriate segue way here I think. Thank you very much.

Agenda Item: Tumorigenicity and Oncogenicity Issues

DR. PEDEN: Good morning in case you have not noticed I am not Andrew Lewis and no amount of lipstick on Keith Peden is going to make me into Andy Lewis.

I am going to present two talks on the two aspects that have been discussed on the tumorigenicity of cells and what information we can get from that information and the second part issues associated with DNA.

We view this presentation as part of our process to educate where we were in our thinking which changes over time. This is part of our continuing process to educate the committee and the public on these issues. What we are going to talk about first is the neoplastic cell tumorigenicity phenotypes and why we want to study tumorigenicity. Then we move on to some of the characteristics associated with the MDCK cells and some of our own data will be on the table. I will say at the outset that these are unpublished information and the lines that we use are not the lines that MedImmune have used so these are ATCC MDCK cells. And we will just make some comments on the tumorigenicity studies that you just heard and some other observations.

So mammalian cells have been used from the '50s on, and the initial cells where primary tissue cultures to dry cells made from the Rhesus Macaque, and African green monkey kidney cells and that was the polio vaccine. There was a big advance made in the '70s when diploid cell strains -- Wistar Institute 38, and Medical Research Council 35, and became used and accepted for vaccine production and as you know there is a very good safety record for vaccines manufactured in those(?).

There was another quantum leap when immortalized cells started to be used and these are Vero cells which are immortalized or what we call neoplastically transformed, but they are nontumorigenic at the passages used for vaccine manufacture. We came to the VRBPAC in 2000 to discuss new lines and these are the 293 which are in fact a neuronal drive line that is not commonly known and PER C6 which is also a neuronal derived line that has been immortalized by the Ad 5 E1 region.

We came to you to ask whether they could be used for the replication defective adenovirus and you said that that was okay and the testing strategy that we came to you with some of which MedImmune has just enunciated, was agreed upon. More recently 2005, VRBPAC we used a highly tumorigenic MDCK cell line for the inactivated flu.

So why do we care about the tumorigenic phenotype of cells? Mainly because we just do not know if the tumorigenic cell poses any risk to the product and therefore the vaccine recipient so we just do not know. Extensive characterization of tumorigenic cell substrate using a vaccine manufacturer's required reagents used in the vaccine manufacture but be well characterized so that is part of it. As we in the division like to talk about it how do we do this tumorigenic cell substrate it is not really understood how to do this but we do the best we can.

We need to have a few definitions because we have used words as you see here rather precisely and MedImmune in fact did define them. Tumorigenicity is really a cellular event. So the process by which neoplastic cells growing in culture form tumors when you inoculate those into animals. The tumors phenotype is an expression of this phenotype genetically cellular traits allow cells to grow in culture and form tumors when inoculated in animals. So that is the tumorigenic phenotype of the cell.

Oncogenicity is an important distinction from tumorigenicity because we are using it to say that these are the agents all clinical viruses that convert normal cells to neoplastic cells and are capable of tumor formation. The important thing about this is based on the question that was asked earlier about rodent cells -- why do we care about whether it is a rodent cell tumor as opposed to the dog cell tumor? It is very important. If the tumor in the mouse or the rat, is derived from the host that says that there are agents in that cell moved from that cell into the rodent and induced the tumor formation in that rodent. So it is a very important question and it is a very important answer to get. So that is why we need to know whether it is a dog cell tumor or a rodent cell tumor. You see what I am saying because that is very important for that definition.

Inoculated cells grow into cells in tumors and like I just said this is the oncogenicity that if you are worried about something in those cells it is very important to determine the species of the cells. Tumors of any mass that develop at the site of inoculation of tissue growing cells when you inoculate them and you really do not know it is assumed or until you do pathological analysis and histopathology so you need to know what was at the site of inoculation, what those methods were.

Tumor aggression is the reverse of a persistent expansion of a mass at the site of inoculation leading to reduction and disappearance of the mass. The tumor producing dose at end point is a way of comparing different tumorigenicity phenotypes of different cells. We are fortunate in the audience today to have a person who knows more about this than anybody probably apart from Dr. Lewis, Dr. Cook right here. So any questions on that I will certainly just refer it to you.

The current perceptions of risk associated with neoplastic cell substrate ‑ neoplastic transformed cells can be nontumorigenic, weakly tumorigenic, or highly tumorigenic. Nontumorigenic immortalized cells are perceived to represent low risks for the transfer of oncogenic activity. Weakly tumorigenic cells require large numbers of cells to form tumors such as PER C6 and 293 and are also perceived to pose low levels of risk transfer of oncogenic activity. Highly tumorigenic cells require many fewer cells sometimes even down to below ten to form tumors and are perceived to represent higher levels of risk. The word is perceived and until we get data we really do not know if these perceptions are scientifically valid.

With respect to some of these risks there is an infectivity risk associated with tumorigenic cell substrate and high instance of viral contamination is pretty much endemic in many labs I am afraid and there is the long histories of the cells that that is a concern that if a cell has been around for a long time it may have been infected, contaminated by viruses that are inapparent just for normal passages.

Oncogenic virus in cells are presently derived from mammalian tumors, for example retrovirus which is the human T cells the lymphoma viruses. Even the human T cells lymphoma viruses, EBV, human herpes virus 8, papilloma viruses are widespread, hepatitis B virus, and of course polyoma viruses certainly are oncogenic viruses.

In addition, it has been reported of lythositic choriomeningitis virus enhanced the tumors and also in human tumors in cell line contamination also transferred to humans so that is again a lab problem.

Then of course what we have described is the DNA infectivity activity retroviral DNA and cellular DNA contain HIV genomes can be infectious at low doses.

Now the oncogenic risk as I am going to go into a bit later is clear that there is a potential for oncogenic risk because we are showing that the h‑ras/myc plasmas can induce tumors in mice and as MedImmune just pointed out our value is now down to one nanogram.

Plasmas with single activated oncogenes ras can induce tumors in mice as well so that you do not need two oncogenes, cooperating oncogenes in fact ras alone can do it. That was quite a surprise to us. It is also in the literature there are about 12 pages that show that tumor zenograph can in fact transfer oncogenic activity cells from the host and that is determined in at least one study showing that the DNA gets transferred. So DNA is oncogenic under low circumstances as well.

When you are doing a tumorigenicity assay you need to select many things. One is the host and what most people use are immunocompromised hosts such as the nude mouse which has defective T cells and so it allows zenographs. And adult mice and newborn nude mice are used. The newborn mice is in the literature we are showned that it is more sensitive as an indicator of a tumorigenic phenotype than the adult mouse. You need to select the dose you can give single dose assays which give you a plus minus answer it is tumorigenic or it is not, but the dose response assay can give you more information and I will say more about that in a minute.

The route of inoculation subcutaneous disease is the most common for tumorigenicity studies but others have been used. The observation period now we are asking for an extension for longer than what is traditionally done so now it is up to four or five months.

The end point selection it could be tumor incidence, it could be tumor latency, or it could be tumor size, weight, and volume. So there are a number of things you can decide to do. Tumor incidence is again the most common one used.

What we are recommending for evaluation of cell substrate tumorigenicity. We are representing a dose of spawn(?) tumorigenicity assay and to save money in animals we are representing every second log. Ten to the 7th, 10 to the 5, 10 to the 3, and 10 to the 1 for mouse, and at least ten mice per dose and record the incidence and size of the palpable tumors at weekly intervals for about four to six months. And necropsy all mice that have died during the study or are sacrificed at the end of the study to obtain any information about what has happened to those mice during the study. We need to do that because it is sometimes you see tumors that exist elsewhere from the site of inoculation.

Here is an example of several data that were culled from the literature, usually from Dr. Lewis's data but they represent separate studies but because of their reproducibility we can I think quite validly put them on the same graph. In green is a 293 cell which is what we decided before was a weakly tumorigenic and as the development of the TPD50 comes down and reaches about 7.5 down to 6.5 down here, it shows you that it comes down over time and reaches a plateau that is a weakly tumorigenic cell. In blue is the HeLa cell which most people would think is the highly tumorigenic cell but again the TPD50 is around five. This cell which is BHK‑21 is highly tumorigenic even more so and is about between log two and three. So this is the way of characterizing the cell phenotype.

Using these analyses you can determine what happens to your cells. For example, if the tumor incidence changes from the expected it could be due to contamination of cells, whether viruses or bacteria. I am going to show you some data on that in a minute.

The infection of the tumor challenged rodent in the assay. So if your animal colony becomes infected and that does happen that can also influence the outcome of your tests done and also the level of course in your confidence of the host adults versus newborns. Sometimes you see it in newborns but you do not see it in adults.

Here is an example of what I was saying about the virus infection. BHK‑21, HeLa, and the melanoma line, 14 out of 14, five out of five, and 21 out of 21 but if you infect these cells with VSC, vesticular stomatitis virus, monkey virus, influenza virus then you can see that now the tumor incidence disappears. This is just a way of demonstrating that in fact the health of your cells is very important for the outcome you get. So if you do not get tumors and you expect to get tumors there is something wrong.

This shows for example here the colony was infected with a very common pathogen the mouse colonies the mouse hepatitis virus and again 21 out of 21 gave it in a normal colony and in an infected colony it goes down a little bit to five out of eight. So these are the kind of data that lead us to conclude that these assays are very useful and they can indicate levels of contamination of either the cell or the animal.

What do they really tell us? Why do we ask people to do that apart from the characterization? Well here is a summary of that again. We can find out what the phenotype of these cells are and then from that the levels of perception of risk are adjusted accordingly. The data aberrations may be indicative of cell substrate contamination as I have just shown you and the important thing about this is it can also identify the presence of oncogenic activity if you term the species of origin of the tumors that arise in those cells.

Moving on to the MDCK cells I know that MedImmune gave an oral history about this. These cells are isolated from the kidney of a normal cocker spaniel in 1958 by Madin and Darby at UC Berkley. They claimed it was a male and I will get to that in a minute. No published record was ever made on the establishment of this cell line. Unfortunately it was never written up as a publication which is you know sounds very less than ‑ for those of us who take a long time writing papers up.

In '64 these cells were at passage 49 submitted to the forerunner of the ATCC, Health, Education, and Welfare, I am old enough to know that. But again there was no published record of the first 49 passages of this cell line.

In 1966, Gausch, at the University of South Dakota obtained some cells and then passaged them in his lab for many years and in fact these are the origin of the cells that Chiron-Novartis used for their vaccine manufacture. Then it became the confusion of the sex of the donor -- as I say it was reported as originating from a male cocker spaniel but then it became a female cocker spaniel in 1994 in the ATCC registry. We can tell you without doubt it is not a female cocker spaniel, it is a male because most of us know that because we have a Y chromosome this animal does not have a Y chromosome, it has one X and it is very unlikely that it lost its second X but it is very common to lose in culture the Y chromosome. It almost certainly was a male cocker spaniel so you can take that as a record.

Now these cells we can talk about that more if you like these cells are a very strange phenotype. They were documented to be polarized cells, they are renal tubular like epithelial cells with secretory, published by Leighton. Some of the phenotypes I think, can be explained by the phenotypes but we can discuss that if you want.

When you look at the literature and Andrew Lewis was the one who scoured the literature to find papers on tumorigenicity, there really are very few papers on this subject. Most of the papers show that they are not in fact tumorigenic. So for example in this case these cells were tumorigenic in chick embryo systems. Stiles had a couple of papers showing that these cells were not tumorigenic in adults but they did form tumor masses in newborns but they also regressed so one showing again that newborns are more sensitive to detect tumorigenic cells but in fact they still regressed.

Darfler showed that no tumors were found ‑ there was no indication of where these cells came from and Kadano got cells from the Japanese Research Institute and they were not tumorigenic in Athymic nude mice. So there are really very few data on tumorigenicity studies of these cells.

In summary of what happened in 2005 Solvay came to us and to you, and talked about their cells but the cells they got from ATCC CCL‑34, they were reported to be nontumorigenic. Then they adapted them to serum‑free medium, expanded them to serum‑free medium to generate master cell banks and end of production cells. These cells were tumorigenic high doses but they had a strange phenotype that they found some regression. This was the first evidence that we had seen that the cells regressed until we went back to the literature. So these cells are tumorigenic but they do regress, some of the tumors do regress.

Novartis Chiron came to us with an even worse line in the sense that we could not understand it. Many people looked at these data and we were just amazed because these cells are grown in a serum‑free medium and they are in suspension, but at different doses for example I think Jim probably remembers at ten cells you have got tumors and at ten million cells you have got tumors, but you never got a 100 percent of the animals coming down with tumors at any dose.

I think most people could not understand those results and came to decide that maybe we needed some Monte Carlo -- I do not mean the gambling -- it is a mathematical model to explain this but nothing was done on that.

The MedImmune cell bank you have just heard about this also came from CCL‑34 and they went through a limiting dilution serum‑free medium and they expanded it one clone in serum‑free medium and passaged this and these are the cells that they are using for the manufacture of their product. These were analyzed for tumorigenicity and oncogenicity in the presence of adventitious agents.

Here is a table that came from their results and since MedImmune had just referred to this table in a far better presentation extracting from the PDF and since I cannot even read it over there either. The upshot of these is this is an experiment done in adult nude mice and as MedImmune just told you there was one tumor right here in a 10 to the 5 dose but that turned out to be a tumor that was not related to dog cells. When they did the study in newborn nude mice again these cells had no tumorigenicity and here we have no aberrations of the finely spontaneous tumors.

When we evaluated the data we were quite I think we were reasonable convinced that they were non-tumorigenic, however there were some things we note in our views. There are palpable masses formed in the adult nude mice but these regressed over time. There was unexpected weight loss during the assay in the adult nude mouse and there were persistent one to three millimeter palpable mass in newborn mice that could not be found upon the necropsy. There are summaries of the results.

I just wanted to tell you a little bit about what we are doing with the MDCK cells and why we are doing it. It really was stimulated by the 2005 VRBPAC and finding the odd phenotype of the Novartis particularly MDCK cells that were unexplained. We thought we should see whether these results are typical or whether they are just specific to the Novartis cell banks and also whether they represented a risk. That is how we get involved in this and as I said earlier these cells are not what MedImmune are using these came from the ATCC as did theirs but at a different time.

What we found these cells from ATCC are tumorigenic and we would say based on the numbers here they are highly tumorigenic. There is no way to get over that 10 to the 7th, 10 to the 6th, 10 to the 5 in adults you can see significant numbers of tumors close to hundred percent. In newborns ten, seven, six, five, four, and three significant numbers of tumors so we really have no choice but to say these are pretty highly tumorigenic.

When we calculate the TPD50 it is 10 to the 5.3 in adults and 10 to the 5.8 in newborns. The tumors are cystic adenocarcinomas based on the histopathology and they seem to contain both epithelial and mesenchymal cells. These are dog cells by carrier type.

During our studies we found some strange things that went on with these cells that many of us had never seen. Not just a mouse but other people Solvay got cells CCL‑34 from ATCC, MedImmune got cells the same cell bank from ATCC CCL‑34 and they were nontumorigenic. We get them in 2004, and they are tumorigenic and we have got three independent vials from different lots, they are all tumorigenic and I will show you those data in a minute, at different levels. Other people on the campus who got cells from another group on the campus and they also got them from ATCC -- they are tumorigenic. That is rather an odd finding.

We can get certain regression of the tumors as well. We see regression in some of the mice during the first eight weeks of in vivo growth. It is hard to explain. Development of cell dose dependent systemic disease syndrome in newborn nude mice and I am going to mention that in a minute what that is. And failure in our hands to find clonal populations of tumorigenic cells that express a tumorigenic phenotype even though the parent does.

Here are the three lines that we obtained from ATCC this blue line was the first vial that we got and this is a replica of four experiments and this one is ongoing so it stops right here -- it has not finished yet and this other one. But you can see right away this yellow line, which is other line which is amp 2 is more tumorigenic than the amp 1 and probably more tumorigenic than amp 3. We have just plotted the Solvay data on here because there were no tumors so it is greater than so we just did that to indicate no tumorigenicity. So that is the line that you would find greater than zero out of whatever mice it is so it is greater than 7.5. This is a result that was not expected.

So tumor regression. What we have here these are our data in this row. We used mice from many different nude mice from many different sources in fact we were also looked at BALB/c nude mice as well. In the initial week we could find 106 out of 120 animals in week two and then 99 out of 120 at week 26. So some animals are regression and the regression incidence here is 7 out 106 so it is not a huge amount but it does regress and that never happened with say for example HeLa cells or other tumorigenic cells. So this is a slightly novel phenotype and I say Solvay reported 23 out of 23 animals regressed. In the literature Stiles found 16 out of 16 were in adult nude mice again using BALB/c nude mice and then by week 26 they had all regressed. This is a study in newborns I am sorry this is not right this is in fact newborns rather than adults and 7 out of 7 in that also regressed. It has been seen in MDCK cells if you look for it in the literature in certain amounts but ours if they are regression it is probably much lower than theirs.

This is an interesting question what we found when we inoculated 10 to the 7 cells into newborn nude mice we had this systemic disease which is we got little tumors, smaller than eight millimeters and they were first detected around 7 to 13 days in 15 of the 34 pups. It does not happen all of the time so this disease is very strange weight loss leading to failure to thrive, emaciation, and death. The first detected around 13 to 38 days when they died and that was 27 out of 34 eventually died. One of the clearly clinical syndromes is kyphoscoliosis which I found out yesterday that it is twisting of the spine this way as opposed to this way but it is twisting the spine.

Our NIH pathologist said if you look at the sections in this you can find invading cells that push against the spine and that describes what is going on so very interesting. We are amazed because again Andrew Lewis in all of his years of experience and maybe Jim can comment either he has never seen this. It is a very strange condition. It does not happen at 10 to the 6 it only happens at 10 to the 7th.

Then what we wanted to do is in fact repeat what other people have done is to see whether you can clone out from your parent population clonal lines that are tumorigenic or nontumorigenic and in fact most people can do this very easily. For example the BHK‑21, melanomas you know when you clone them you get four out of four clones and they work fine. When we come along to ours which is on the one line here we have 15 clones now tested where we have other clones. Only one of them caused induced tumors and not all of the animals did. So far 75 mice we have tested with these 15 clones only one has come down with a tumor. So in 14 lines there was not a tumor at 10 to the 7 cells, and in one line occasionally a tumor came down. This is again different from the standard tumorigenic cells when you clone them you get clones that are tumorigenic.

So in summary of this part of the talk, the limited information regarding the origin derivation of passage history of MDCK cells. Inconsistencies are in the reported phenotype of these MDCK cells, for example, the Solvay are tumorigenic and Novartis are tumorigenic and nontumorigenic MedImmune phenotypes. And cells using dose response, the tumorigenicity assays have turned out to be surprisingly complex.

When we started this say in 2006, we thought this is going to be very straight forward and it turns out for all of those reasons we just went into, it is very complex and we are still not sure what is going on.

Dose response tumorigenicity we feel are valuable because they can provide information to differentiate the different tumorigenic phenotypes and insight into the possibility of cell substrate contamination. The role of passage history adaptation to serum‑free medium and cloning of the tumorigenic phenotype and the behavior of the cells in dose response tumorigenicity assays are not always clear.

The so the people involved in this of course this work is being directed by Andrew Lewis, he and I have been long time collaborators but this really, the tumorigenicity side, is directed by Andrew and most of the work was done by Romelda ,who is here, and Gideon is our technician, and Lauren Brinster who is the pathologist at the NIH.

So I will stop right now before I move on if there is any questions of this part of the talk or should we go on?

DR. MODLIN: Let's go ahead and take some questions if there are for Dr. Peden specifically for this part just a very, very quick one go ahead Bruce ‑

DR. GELLIN: You mentioned at looking at the duration of observation you extended the period from I forgot three months to six months?

DR. PEDEN: In fact sometimes it was even 21 days ‑

DR. GELLIN: How did it come up with the numbers and what did you accomplish?

DR. PEDEN: Well it is a balance between having assays that go on for the life of the mouse which are expensive and what do they provide you versus the ability to detect tumorigenic phenotype of a cell. So it really is based on the VERO cell. Our work with VERO cells, they took much longer to develop and within we felt five months or so, you could detect all of the VERO cells that were going to come down. At least say 95 percent of any tumors. So that is really where it came from. It is a practical consideration versus if you are a scientists you would put it on for the life of the mouse but that is not the practical thing so really it is a balance between trying to detect it and cost and time.

DR. MODLIN: Anything else Bruce?

DR. GELLIN: I just wanted to ask a broader question in the background materials I felt were really very helpful and there was a comment in there about how this may all depend on genetics. How do you put all of this in perspective of these nude mice and various animal models and what the relevance of that is for human vaccine?

DR. PEDEN: Oh, what a terrible question. We labor over that. Historically, FDA has asked and other regulatory authorities have asked is your cell tumorigenic? What we do not know is does it tell us anything? I can just showed you what it can tell you but is it relevant for a safety of a product(?) we just do not know. At this stage we ask for it because it is the best we have and we really need to know if they are tumorigenic, down the road we need to have that information I think. But what assay used exactly because there is a difference between athymic nude mouse and a BALB/c nude mouse uncertainty then you could go to nodskid(?) mice which we were now doing for oncogenicity studies, is that better and if it gives you a tumor in say a non-skid and it doesn’t in the nude A5 AG(?)how do you balance that risk assessment? I do not know. I think we have to do certain kinds of experiments to try to get at that but I agree with you it is a very complicated issue and at the moment we are just asking for the athymic nude because most of the ‑ it is a widely available animal, it is cheap, and most cells that will form tumors form tumors in that animal, but not all.

DR. MODLIN: Could I further confuse this picture by pointing out that all of these assays are done by injecting these cells into the skin or under the skin. We are talking about a vaccine that gets squirted into the nose here. Has anybody once taken any of these cells squirted them into the nose of any of these animals and see what happens?

DR. PEDEN: I do not know whether MedImmune has done it we certainly have not done it. Do not forget these are historical. I agree with you completely and it has certain relevance not so much for cells since this product has cleared the cells but it does have relevance to the DNA. So I absolutely agree with you.

DR. COFFIN: There is a little bit of discordance between some of the things that you said and some of the things that the MedImmune presentation had. Regarding oncogenicity, you stated that when you inject cells and see tumors and at other sites that is a sign that there is some agent or DNA or something being transferred from the donor to some other place in the animal causing a tumor there. These several tumors of that type that were in the presentation were all dismissed as being background.

My question is do you have a clear definition of background in the system and what above that you would become concerned that there was some oncogenicity effect due to something being expressed by the cells that were injected?

DR. PEDEN: That is a very good question. You determine that the cell is crossed with a dog cell and it has migrated away that is clear. The question is when is the cell or a host did it arise because of something in that cell had to have transformed the mouse cell a cell of a mouse or a rodent. The only thing that we can do at this point is to ask for repeater sequences that they are present. So we only asking for whether the DNA is taken up by those cells and if it is then these repeater sequences will be present, we think. Now if they are not then you are never going to find it and if it is a virus you are not going to find it either this way so you just do what you can do. But I think what you just said too it is not the frequency of the event. And also the other thing we need to know is what is the background frequency of these spontaneous tumors in that strain of animal that we are using?

DR. COFFIN: What is the standard by which you a assess something as being enough over background to express concern?

DR. PEDEN: I think if the, maybe I should not be answering this, but I think if the test article and the controls have a similar frequency I think ‑

DR. COFFIN: But we did not actually see those data in the MedImmune presentation. We did not see uninjected controls for example, age matched, litter matched, whatever controls that would show the same background of these other tumors. There was a statement I think just based on sort of a vague statement based on perhaps historical maybe we need the MedImmune people to address this more clearly based on historical controls or based on what is in the literature which may or may not be appropriate to the immediate situation.

DR. MODLIN: It is perfectly fine if you want to address that. Please do, Dr. Kemble.

DR. KEMBLE: In the tumorigenicity studies ‑ so that is where we saw the histiocytic sarcoma and yes, two points one it was one animal out of that group and it was not dose dependent. You are correct that background rate was taken from historical literature where you can find histiocytic sarcomas in these animal species it was not a matched cohort.

In the oncogenicity studies we did have the PBS and non-injected routes where we did have some background rates as well. Were you talking specifically about the tumorigenicity studies, I guess was my question.

DR. COFFIN: Well, yes I was speaking to the point that Keith had made about seeing tumors at other sites when you inject cells. So yes the context of tumorigenicity studies.

DR. KEMBLE: I think that we can certainly state that within that strain of animal we have some data but certainly not within a matched cohort born at the same time in that same study.

DR. MODLIN: Thanks Dr. Kemble. We do need to move on. I guess the one take home is all MDCK cells are not the same. I think that we have learned that but beyond that it is a matter of speculation.

DR. PEDEN: But they all came from the same progenitor so yes you are right but that is really ‑ you have to extrapolate what has happened to these cells. It was at a function of the original cell the phenotype and then all of these phenotypes are there in some way but different subclones, different things that is true.

DR. COOK: There is another take home though, I think if your data on subclones is correct it suggestions that the majority of the MDCK cells are not tumorigenic no matter where you get them from. So if you go to the ATCC and you get different vials and you subclone them and only one out of fifteen subclones turns out to be tumorigenic that holds up. That suggests that the majority of the population of those cells grow from ATCC sources no matter how long their passage history, if they were to be subcloned would not make tumors. Am I correct in understanding the data?

Because one of the problems that I had is trying to understand the difference between the MedImmune data and the Solvay data and everybody else's but now that you have done that experiment that suggests that it is possible that MedImmune could have cloned out a nontumorigenic subclone with MDCK you must have got it from your same vial.

DR. PEDEN: It is certainly possible yes we agree. We were not expecting that result I mean we just did not know but yes I agree with you completely Jim.

DR. MODLIN: Let's go on.

Agenda Item: Issues Associated With Residual Cell-Substrate DNA: An Update

DR. PEDEN: So my second charge is to talk about issues associated with DNA and this all came about because the sections of risk of DNA vary from ‑ it is an impurity to the amount that needs to be measured but it is not a safety concern, bearing to its biologically active molecule whose activities pose a significant risk of acting recipients thus we have got to limit the DNA and its activity.

You can imagine that both sides of this spectrum existed in everywhere you can look including in the FDA. We decided that we needed to understand whether the DNA is in fact biologically active at the levels that we have used and so what we wanted to do is to develop assays for both the oncogenicity activity of DNA and its infectivity activity.

These are the major activities that could pose a risk in a vaccine. I have deleted any reference to integration and we can go over that if you wish but integration has been considered by many committees both for the WHO thus we have got to limit it and its activity. The WHO and in fact even this committee that because of the inefficiency of DNA integration per se and certainly inefficient of integration in the type next to gene is so remote that integration has not been considered a significant risk. So the risk of oncogenic activity is not from integration and activation it is due to the introduction of an oncogene, an activated oncogene to convert the cell directly. The infectivity activity is to generate an infectious agent the DNA virus or retroviral provirus.

So I will just remind you that the WHO recommends DNA limits for parental vaccine is that when they looked at this they decided there should not be any limits for primary cells or diploid cells, but make it since at least 1996, I think make it less than 10 ng per dose. Why these primary cells there was no limits I guess it was maybe due to chicken cells but there were not any limit.

How do we resolve these issues? Well because there were no data in the literature or very few we decided the only way to do this is to find out what the effect of biological of DNA at what levels. So we have been developing sensitive and we hope quantitative assays to measure the biological activity of DNA use these data in the form of these sensitive assays and estimate the risk of a particular event, a DNA event. Use assays to quantify the amount of reduction of biological activity afforded by particular treatment for example the nuclease digestion that you have heard about also chemical inactivation. Then try to use these data to try to estimate safety factors for a product with infected DNA so that is our approach.

What we ask people to do and this was really initiated in the 2001 meeting with the PER C6 DNA, is we asked them to clear the DNA as much as possible. So this clearance is reducing the amount of DNA to less than or equal to 10 nanograms per high end dose and reducing the size of that DNA to below 200 base pairs. This is what we asked people to do.

Now we recognize that in some live vaccines it is not always possible to reduce it to below 200 base pairs. We felt at the agency that reducing DNA below 200 base pairs probably at least at that time will consider that you removed any biological activity of DNA. But again as I say we understand that not every product is hearty enough to be able to get the DNA below that so we are not unreasonable.

Evaluation of the cellular DNA, now we want people to inoculate 100 micrograms of cell substrate DNA into newborn hamsters, newborn rats, and newborn nude mice these animals are the most sensitive to detect oncogenic viruses. Then we monitor the animal for five months for tumor formation and general health and determine the species of the tumors that arise. You can differentiate between spontaneously arising tumors and tumors as a consequence of the DNA.

Now I have to mentioned of course that we also ask for cell substrate, yes lysates from the cell substrate but this is a DNA subject talk so that also is here. When we brought it to you, you were very wise in 2001 you said what the hell is this assay? What will it tell you? We discussed it and you agreed that it probably would not work but it is the best that we have. It is the best at the time so we need to do it and so that is how it was left in 2001. There was no evidence that it would work but it is the best we have. I am going to show you some data that say in fact maybe it will work.

So why can't you just test the DNA as we were saying? What is the problem? Well the problem is the size of the genome. The haploid mammalian genome is about 3 times 10 to the 9. I know the dog genome is 2.49 or something but it is about 3 times 10 to the 9 from the mammalian genome. However single copy genome virus, and I just used three because it is arithmetically easier, 3 times 10 to the 3 times 3 times 10 to the 4, 3000 to 30,000 let's just assume that. If that is true, therefore the single gene or virus is about 100,000 to a million fold less abundant for the equivalent amount of cellular DNA compared with the plasmid DNA containing the same genes or virus -- just mathematics.

That is, and this is the important thing, the amount of cellular DNA equivalent to one nanogram of a cloned gene or virus – if that is what is active biologically, is now 100,000 to a million nanograms which is 100 micrograms to 1 milligram of DNA. I had this similar slide in 2005, we had numbers about a little bigger than that so I had up to a gram of DNA - you may remember. But now, in fact it is more realistic because of our data, one nanogram might well be biologically active. It is bringing it to 100 micrograms to one milligram. Which is getting into the range where you might be able to detect it. The other problem is there is no standardized assays to do this.

So the oncogenicity testing of the lysate by MedImmune, again you have just seen this. They inoculated lysates into the animals that we asked newborn nude mice, newborn hamsters, and newborn rat, and they found as you just heard, that the assays did form tumors but they were reasoned to be spontaneous and not related to the test article by illustrating there were no dog sequences present in those tumors. Now we have just discussed the possibilities of that a moment ago.

Here is the DNA results again similar results 100 micrograms of DNA in the three assays the three animals and again there was no as we could see no specific tumors arose as a construct of the test article. In these tumors we think that the assay have showed that there were no DNA induced tumor events because of the canine sequences that were absent from those tumors so we accept that.

The risk estimates you have heard although they have adjusted them slightly because of the package that we gave them about one nanogram but the safety factors they estimated them about 6 times 10 to the 9 for oncogenic events and the risk for a single dose is 1 in about 10 to the 10 doses. These could be adjusted based on our data that I am going to show you about now that what we did is put both oncogenes on the same plasmid so this is in the other you have got a yellow oncogene called myc and a red oncogene called ras and they are driven by the Moloney sarcoma promoters and they are in separate cassettes so this is our what we call the dual ras/myc expression plasmid and Dr. Hughes and Dr. Coffin will be very aware of this plasmid.

So this is what we chose to inoculate into mice. I am going to show you now that these are fairly new data. We wanted to find out how sensitive we could ‑ how little DNA we could detect so in fact we initially used circular DNA and in fact the WHO last year I showed you we had 25 nanograms were oncogenic and that is this point right here. Then we decided that we have got to look at linear DNA because there are some reports in the literature that linear DNA is in fact more oncogenic in the polyomavirus field and also as we know from our in vitro sometimes it is more active in vitro. So we did that and low and behold we can go down to ‑ first linear DNA is more active than circular but now it is down to 800 picograms were in fact oncogenic and induced tumors and these are not just one tumor these were several. So this has been repeated and this is an average of two experiments and this is an average of about three or three and a half experiments here to give you these lines.

Now we are talking about less than a nanogram of DNA is oncogenic. But to make that assay work and this is the assay that John asked about in fact, how do you a assess that the oncogenic activity or with cell substrate DNA over 100 micrograms of lysate from 10 to the 7 cells is it meaningful to do that? In other words if you had an oncogenic in there could you detect it in the presence of this vast quantity of normal DNA or normal lysate and we have done some experiments now on normal DNA.

What we wanted to know is to determine whether nononcogenic activity in large amounts inhibits the detection of an activated oncogene. The experiment is to keep the positive control DNA constant and this was about two micrograms of a linear oncogenic DNA and to see whether you increased the amounts of normal mouse DNA does that excessive nononcogenic DNA interfere with your ability to detect oncogenic activity.

As you can see in this graph right here this is a percentage on this axis on the Y axis of mice with tumors. In this case it is going close to 100 percent and this is the amount of non-oncogenic DNA normal mouse DNA, 100 times 50 and 10. What is absolutely clear is that up to 100 micrograms of DNA does not obscure the ability to detect oncogenes at least at that level ending. Now this result that very surprising to us I did not think it would happen but again I think Steve may remember when we talked about this at the WHO a year ago and this experiment really shows that the recommendation for 100 micrograms might well have been perspicacious and not completely stupid although we do not know that yet.

The summary of the oncogenicity results the CD3 epsilon mouse which is the mouse that we are using for this appears to be the most sensitive mouse to detect oncogenic activity. Linear DNA to under a nanogram is of the positive control DNA which we have now can be detected. Cellular DNA after 100 micrograms does not appear to the oncogenicity of the ras/myc. I think the oncogene I have not shown you the data for this but a single oncogene seems to reinitiate the process and subsequently takes longer for tumors to come down and one of the first ones we found it takes eight months before the second event or whatever the event is to occur without a frank tumor that contains the ras DNA so we are now part of our research program is to try to identified what that second event is and so far we have had three examples of ras alone inducing this so that is quite interesting.

I think in conclusion of the oncogenicity part we can say the DNA appears to be more of an oncogenic risk than was reported or even realized. The CD3 epsilon mouse maybe sensitive to assess the oncogenic activity or cell substrate DNA. We are in fact doing experiments right now with MDCK cell DNA with a positive control plus or minus and they will be our recommendations to testing 100 microgram cell substrate DNA seems to be supported by the data and also indicates a need to include a positive control in this assay which we are developing.

Infectivity. Why would anybody study infectivity? Well it turned out that infectuary risk may have been higher risk than oncogenicity risk for DNA and in fact this was discussed I think in the 1998 VRBPAC and they wisely said that this might be important and I think I cannot be sure but I think John Coffin was on that committee. I many not sure about that but I think he was one of the people that thought that the infectivity risk might even be more of a risk because of course once you have infected with a virus then you can amplify that in vivo. That is one of the reasons why it is clearly an important consideration.

Specific infectivity of DNA for different viruses is just unknown. You look in the literature there are no people who have really looked at different types of viruses for their activity and so that is one reason we needed to get some data on how infectious DNA is.

If we use that assay and those numbers we are also unclear the DNA oncogenicity of course because if the assay is more sensitive than anything that clears the more sensitive product you would clear also the less sensitive product in this case DNA oncogenicity. Also more importantly it will allow other aspects of DNA activity to be studied so I will talk about that in the end.

What we found over the years is that one picogram of HIV DNA clone is infectious in vitro we can detect that in our transfection co-culture system that I have presented I think in 2005. That corresponds to about 100 thousand molecules so I think that is important to think about the number of molecules that this represents because if you start clearing it by 10 to the 7, 10 to the 8th then almost certainly you have cleared every single molecule that have been there.

Also now we can find two micrograms and the data that we had before was 2.51 microgram of this particular HIV infected DNA is not infectious but two is two and above. So these are the limits that we have so far.

DNA inactivation measures importantly for an inactivated vaccine you have many more opportunities to kill DNA and also adventitious agents because these compounds B‑propiolactone, formaldehyde, and binary ethyleneimine which I used to inactivate viruses. They also effect nucleic acids so this is a very important second tier approach and what most people knew that you have heard today is the bacterial nuclei benzonase can only be used with live actions because you of course cannot inactivate the vaccine. So this is the only line of defense really is nucleic digestion and clearance of the mouth.

Here is the experiment that I was talking about. What we do is we mix the clone DNA with jerkout(?) DNA and digest it with benzonase for different amounts of time, going up to 15 minutes at 30 degrees or so but any ways just to get a nice distribution. Sso maybe you mathematicians in the audience can tell whether this is a Poisson distribution or not I do not know. What we did is mix equal amounts of jerkout DNA with the HIV clone and then purify the DNA from each fraction and then transfect it in our assay system, transfection co-culture system to measure the infectivity and determine at what digestion point infectivity is lost. Now of course as John Coffin pointed out in 2004 at the cell substrate meeting this is a great(?) DNA of 10,000 base pairs, but if you look at the infectivity it disappears at a certain point. Surprisingly in fact this one right here does have infectivity so clearly 10KB which is over there which is very important, it may be apropos to what Steve was asking maybe digestion is not random with respect to all sequences but if you go down below 600 then it disappears into the medium size and around 300 disappears into the low. This result you can use 750 nanograms of DNA instead of 150 of viral DNA, so you can use that to estimate risks.

How do we do this? I know the statisticians in the audience are going to hate this but this is how we did it. The factor by which an event resulting from a biological activity of DNA is reduced we come up with a safety factor. So it is just a factor. So if you can remove it by a certain amount that is the safety factor. This reduction can occur by lowering the amount of DNA and or by inactivating the DNA. So therefore it is analogous to clearance of adventitious agents, those two processes, removal and inactivation. Inactivation is by, of course, nuclease digestion or it could be by chemical inactivation.

Now what we think and this was ratified again at the 2005 VRBPAC meeting -- safety factors of 10 to the 7 or more will provide substantial safety margins to DNA activity. You can argue about it but this is what we decided.

Calculations for the sake of time I did not go through how we get these numbers I can do that if you wish or I am sure John Coffin could work it out in his head and tell us all about it with the numbers of genotype but the proportion of the retroviral genome in a mammalian genome and based on say one nanogram of cellular DNA.

Now we know one picogram of HIV DNA is infectious and so therefore based on the proportion of HIV DNA in the genome of say an mammalian cell if you calculate that you only have a safety factor of 600 which means to me and to us it that you have got to do something with the DNA. Because if for example the VERO cell, the MDCK cell had an infectious virus that I were not aware of even though there is no errors for it then a safety factor of 600 it is just insignificant. It is not sufficient in our view so therefore you digest it. Now when you do the calculate highlights that safety in addition you get by digestion that safety factor now goes from 600 to 4.5 times 10 to the 8th from the infectivity assay. You get this extra value of safety by digestion of the DNA and this is based on experimental results and then extrapolating from experimental results.

So for the oncogenicity study we have shown that one nanogram of the ras/myc DNA induces a tumor. Oncogene represents 10 to the minus 5, 10 to the minus 6 of the mammalian genome and therefore that is 10 to the 5, 10 to the 6 nanogram of cellular DNA will be required to induce an oncogenic event and therefore the safety factor for one nanogram of cellular DNA in the vaccine product would be 10 to the 5, 10 to the 6 which is pretty good.

We need to adjust these though. Safety factor excludes that two oncogenes are on the same cell are required we kind of do not like to consider this because our result with a single oncogene says that a single oncogene can induce something that is nasty down the road will ‑ a second, third, fourth event can occur, could occur and so I think we need to use the safety factor based on just one nanogram of the plasma(?) so rather than factor in the fact that two oncogenes are required to go to the same cell.

But we get an additional safety factor from the reduction of DNA. If you add the safety factor which is about 7 times 10 to the 5, now the oncogene safety goes from 7.5 times 10 to the 10 to 7.5 times 10 to the 11. These are real numbers. These are big numbers. However it gets reduced because of the number of oncogenes in the genome. Now there are many calculations about dominant oncogenes about 200 and if we factor in now we know that they are micro RNA genes that are oncogenic, they can effect cellular metabolism -- let's just raise it to 300. This is got kind of a fudge factor but then you of course decrease the safety from 7.5 times 10 to the 10 to 2.5 times 10 to the 8th and 2.5 times 10 to the 9th. Still very large numbers and again we are assuming that even if the genome has 300 oncogenes they may not all be active in this particular system. So in other words some oncogenes only try and form certain cell types but you know this is the only thing that we can do for calculation.

So the root of inoculation is another safety margin because it has been shown in the literature by Mel Martin, Wally Rowe, Mark Israel, their pioneering studies on cloned polyoma virus DNA when they inoculated subcutaneously or IM or variosgot(?) got tumors but not with polyoma virus taken through the oral route. Then the Merke group, Jose Lebron and Brian Ledwith and colleagues, showed that there is about a million fold decrease in uptake if you put DNA through the mouth in a rat versus IM so you get a huge safety factor here because of the decrease stability to uptake DNA through the oral route. What is known in the nasal region is a bit more restricted. It is known that you can inoculate DNA vaccines via the intranasal route and you can certainly get an immune response usually that requires adjuvants which means that the naked DNA itself is taken out very inefficiently and as you have just seen some data from MedImmune and they factor in a value of about 10 to the 5 difference between an IM introduced DNA and from the intranasal route. So whether this is the only number or it needs to be adjusted based on activity of biological molecules we do not know but that is what we have so far.

So what do we need to know? Some of the ones that we need to know is this one. This is a huge issue because anything in the vaccine that is not connected to DNA it is chromatin. It is very difficult to assess the activity of chromatin because the assays are not really designed to it. Plasma DNA you cannot make them chromatin. You can I suppose, by making a nucleic histone version of it.

I think with our infectivity assay in vitro we are in a position to look at this because we can make DNA from the HIV infected cell. So we can make DNA and chromatin and compare the activity of both. So we plan to do this. It is still not quite clear whether the nasal route, if you have a smaller molecule of biological activity where the 10 to the 5 fold is less efficient than IM. We are generating the reagents now to see whether oncogenicity can be detected via that route. We would like to put a reporter gene on that so we can monitor them out so we can see whether the DNA gets taken out. So that is what involved in now.

We still do not really know the frequency of the DNA induced initiation event. It would be nice to know that. Again I think we are in the position to get a handle on that now we have got sensitive mouse models. We can look at that single oncogene initiation event. The other thing that is slightly difficult to get a handle on is whether heritable iatrogenic events can induce oncogenic events in vaccine recipients. Now whether these could pose a safety concern with the description of micro RNA is this becomes a bit more perhaps of a concern than it was several years ago because these micro RNAs are stable and if they can be taken up by cells in vivo and they can induce a change in that cell that is heritable then that might be something that we need to consider. We are in fact starting to address that by looking at micro RNA genes that have been shown to be oncogenic in vitro to see whether they induced oncogenic events either individually or in concert with other oncogenes and perhaps we can look at micro RNA's themselves.

Our overall conclusion based on what we have looked at from a live product MDCK cells the amount of DNA one nanogram per dose as I told you the infectivity safety factor is 600 which moves up to about three for oncogenicity is about 300 to 3000 using the value of about 300 oncogenes. However if we reduce the size of the DNA below 600 base pairs which is what our data have shown you can get an infectivity safety factor now about 4.5 times 10 to the 8 and the oncogenicity is about 2.5 times 10 to the 8, 2.5 times 10 to the 9 depending on what you call an oncogene. In addition if you look at the inoculation route you get another 10 to the 5 fold so these numbers have moved up by 10 to the 5 and so again we see they are rather big. But even if you do not take the route of inoculation into account then I think by digestion of the DNA I think you can see that you get large numbers of reduction in risk.

What do we have for the MedImmune? Their data show that their clinical loss is about 0.1 nanogram. The immediate size of their DNA of 450 base pairs 64 percent is below 500 and 90 percent below 1000 so these are the numbers that they have provided. So this probably is these numbers I think would survive the test for risk assessment so we feel they have addressed the safety concerns with respect to DNA. The people who have done the work most of the in vitro work on this activity all of it was done by Li Sheng Fowler and also the DNA oncogenicity the main person was being other people I know you have got a couple more technicians, people to help and Andrew has done collaboration with Dr. Lewis and as you can see here the couple of people at the table both Steve Hughes and John Coffin have provided very opposing and wise counsel for that and we did get some funding from some sources the Commission's office and the National Vaccine Program Office and the contract from the NIAID that enabled these studies to be done.

DR. KEMBLE: Thank you Dr. Peden again very, very helpful straight forward, clear presentation. I wish the answer was as clear the interpretations. Let's see if there are further questions for Dr. Peden specifically on the oncogenicity data.

DR. WHARTON: For those of us who do not think about oncogenes most of the time when in these models when the mice get oncogene related tumors what kind of tumors are those?

DR. PEDEN: The tumors that we find that the rats make(?) are undifferentiated sarcomas. That is what they are now it is possible that you get other tumors of course with other oncogenes in different class we have not you know we only have limited amount of resources so we have not concentrated on the who will spectrum of oncogenes. It would be nice once we get our system worked out I really want to test a few other oncogenes to see what sort of tumor incidence you get and also what tumors may come so it is interesting but a very good question.

DR. COFFIN: Contrary to what you might have thought I was not able to quite do the calculations in my head as the slides went by but in the slide where you showed the degradation of DNA with benzonase and the infectivity of these spiked HIV plasmid was in fact the cutoff that you saw what you would have predicted from a Poisson distribution given the numbers of copies in which you would estimate for the specific infectivity of those copies ‑ so in other words when you went from minus would you ‑ the last positive would have had at least one surviving infectious copy and the first minus would have had none by Poisson distribution does that in fact follow for those calculations?

DR. PEDEN: We have not done that yet what we are doing now is trying to develop mathematical models to look at that. We have not in fact determined whether it is a Poisson distribution. We are doing that now so we do not know that. I was reluctant to do those kinds of calculations because I thought the data would be better than the calculations but I agree with you. I have come around to thinking mathematics is more specific.

DR. COFFIN: Well the only thing that would be nice to see if there was some indication at point to that digestion was less uniform than you would have thought and that would be a simple way to get that kind of indication. I would not necessarily say it was going on but it would give you an indication that there was something.

DR. PEDEN: Absolutely it is hard to imagine that because we have done this with two different nucleases two different ones with different specificities and this is naked DNA so it is hard to imagine they are will be such a resistance part of DNA but I think your point yours and Steve's about if it is chromatin and cellular material that is a different issue I agree with you.

DR. MODLIN: Other comments or questions? If not we are right on time I suggest that we will break for lunch.

(A luncheon recess was taken at 12:30 p.m., to reconvene at 1:35 p.m.)

A F T E R N O O N S E S S I O N (1:35 p.m.)

Agenda Item: Open Public Hearing

DR. MODLIN: Let me ask Christine if there is anyone who has signed up to present at the open public hearing?

DR. WALSH: No, there has not been anyone.

DR. MODLIN: Is there anyone present who wished to make a statement as part of our meeting. Seeing none we will move on to the next item on the agenda, which is committee discussion. First of all I would certainly like to congratulate both the FDA and the sponsor MedImmune for very clear and focused presentations today. It was very useful and I think the committee has a clear sense of what the issue or issues are. I think we would all agree that there are a number of complexities and that we are certainly operating in a kind of a broad sea of uncertainty as well.

I think the real purpose of the discussion this afternoon is just that, the discussion is probably just as important as the ultimate vote is. I think what I would like to do is open it up. Clearly we are focusing on both of the questions but in many respects the discussion for question two will be driven by what we think about question one.

Agenda Item: Committee Discussion

DR. MODLIN: So I am going to open this up to the Committee. Certainly opinions are important. If there are additional questions that you want to direct either to the sponsor or the FDA please feel free to do so. I think the important thing is to try to -- also see if we can identify any additional issues of importance that Mort raised or in any of the presentations and the questions this morning as well.

I think perhaps the best way to do this is to -- one way would be to go around the table and ask individual Committee members what is on their minds whether there is additional information. I think we will do it in reverse order, and maybe start with Dr. Self. If you wanted to start off and then we can just take it form there.

DR. SELF: Okay. I guess my focus has been on the two aspects of the manufacturing process, the filtration and the digestion. Even though it is mundane and the calculation you described is kind of a standard one, I think it would be useful to hear some more details about the nature of the filtration and the justification for that calculation, because one could easily have a somewhat different model based on pure physical filtration but that calculation is not the one that is appropriate, and would bring the safety factor that you cite down into the range of debate. So getting a little more detail there even though it is mundane.

The other aspect is -- I really liked the data that was presented in the last presentation that shows in more detail the distribution of fragment size for different kinds of digestion, I guess. U was impressed that there was this element of that distribution that persisted for some time at greater fragment size. And I think the risk based on the data that I have seen today is really associated with that element. It is clearly not a Poissan distribution. So a finer characterization of that distribution in the manufacturing process you are proposing I think would be the other piece of information that would be really useful, now if you have it, but certainly built into some sort of monitoring over time.

DR. MODLIN: Anyone else from the sponsors who would like to respond, please do.

DR. KEMBLE: Sure. Let me start with the filtration, again to put it somewhat into a context. In one example, where you had two identify=cal filtration steps serially to each other, then I think you are absolutely right, combining the two the way we have is not appropriate. I can’t remember if I said this earlier -- I may not have -- the two actual filter substrates, the pore sizes are different for the two, and so the way we see it si that first filtration event has a certain power, and then that second filtration event quite honestly is challenged with a significantly reduced burden, if any burden, and so that is why we still feel it is appropriate to combine them the way we are.

The piece that we haven’t calculated -- so if we were to take some of the factor back from those combinations, the influenza virus is a lytic virus. A lot of the cells are lysed before it ever is harvested. There are identity chromatography steps and other steps that we simply haven’t measured into that factor. I think what we have tried to model is something of a conservative number because we don’t really know how to estimate those experimentally right now.

We can try to do that more as we develop the process further and see if we can come up with more accurate assessments, but I think it is still on a fairly high order of magnitude from when we consider all those different factors including the benzonase treatments steps and the viral lysis and identity chromatography.

For the DNA distribution we have only done a limited amount of work doing scans and some relatively more modern techniques to look at sizes. Again we don’t see a lot of evidence for something above a thousand base pairs. To the point made earlier that is why, when we did the modeling, we assumed the worst case scenario where in one case it was fivefold larger and in another case it was over tenfold larger than what we actually have in the vaccine in ten times more amounts.

So we think even if we find a fragment of that size, in real time it will allow us to recalculate the safety factors, but we would have very, very miniscule amounts at that level and it will probably drive the safety factor higher rather than lower.

That said, our distribution -- I could say it is a Poisson distribution because that is difficult to do with these. There certainly is a peak around 450 with tails on both sides and it is very difficult to truly quantitate if that is 80, 90, 100 base pairs at this point. WE are trying to get technologies that do that better for us. We aren’t there yet.

DR. MODLIN: Thank you. Dr. Self, anything else?

DR. SELF: No.

DR. SANCHEZ: My question was also directed to the fragments.

DR. MODLIN: Dr. Hughes?

DR. HUGHES: I would certainly echo what my colleagues have already said. One of the things, to amplify this just a little bit, although Dr. Friedman’s work is lovely and it does speak to the problem to an extent, it is important to point out that he is doing the digestion negative home(?) and that is probably not what is being done in the system you are using. That does not mean it is not going to work, but it is a somewhat different question.

The other thing that could also be said is that it is at least conceivable that a significant fraction of the ten percent of the DNA that has not been identified is significantly larger. That would actually mean that in terms of reduction of the potential pathogenicity, whether it is by infectivity or by oncogenesis, that your safety factor would be tenfold, not orders of magnitude.

I am not saying that I think that is true. You guys have done a very nice job of controlling for most of the things that can be controlled for, so controlling for that would seem to be one of the few things left to really make everyone confident that there are no askable questions that are left unanswered.

The other thing I would say is just to repeat what I said earlier, which is to include particularly in your lysate injections some sort of positive control. Again, this is not meant to be a criticism of what is done, most of which is really very nice, but it is really a matter of completeness. I am not saying I expect that to give you problems. In fact I think it is at least plausible to think that the lysate will tend to break up the DNA and adjust it and you will be safe, but I would be comforted, and I would hope you would be comforted, if you got a good answer to that question.

DR. MODLIN: Dr. Coffin?

DR. COFFIN: I will start by just seconding everything that went before. I completely agree there are positive controls in the assays that would be very desirable, although Dr. Peden did in fact show such a positive internally-controlled assay, which should be very reassuring to you. Also I think it would not be very difficult to at least accurately quantitate the removal of DNA fragments larger than say the 10kb size. I could design a PCR experiment on a napkin here that probably would do that job for you quite nicely. I am sure you can do that, too.

In general I think the data here -- I would answer that question yes, although I would like to see at the current state of the art, but I would like to see some of these other things done a little more along these lines going forward.

We have been asked to set aside the issue of adventitious agents, which actually worries me more than oncogenic DNA, to be perfectly honest with you. I think that is something that needs a little more attention. We are moving into the era, and I am speaking sort of in the future, not addressing this issue, but we are moving into the era where it will be possible to get a complete sequence of the DNA. You have probably considered that. You have probably had these discussions.

It will be possible at a reasonable expense to get complete sequence information of the DNA that is in your cell line starting out, and maybe with that and with comparing some different clones you will be able to learn what it is that makes these clones different. And I suspect that if we were having this discussion five years from now we would be discussing a lot of that kind of data. So it is a little bit of a moving target here, and I think moving forward there will be a lot of that being done.

The other thing is that to my mind could be the most definitive way to identify adventitious agents. You could do a selection experiment and throw out everything that is dog and see what is left in a fairly straightforward way, and not for huge expense these days actually.

DR. MODLIN: That is an interesting comment. Norm, do you suppose the FDA is going to require that all cell lines in use for live viral are going to be required to be completely sequenced? You don’t have to answer that question.

DR. BAYLOR: I will at least comment as Dr. Coffin said - in the future.

DR. COFFIN: Yes, it is in the future. I put this more for these guys than anything else, but I think the adventitious agent here should not really be completely taken off the table even though we have not discussed it. I think it is something to keep in mind and perhaps when looking into the larger issue, in question two perhaps, that would be something to worry about.

DR. MODLIN: Thank, you, excellent point. Seth?

DR. HETHERINGTON: I think as I have heard the discussion go around the table after this morning there are a couple of things I want to point out. First of all, MedImmune laid out an excellent case as to why they are doing this to begin with. There is clearly a medical need here, particularly with the concerns around pandemic flu and the ability of industry to respond with adequate amounts of vaccine in a timely manner. This is a new technology that is clearly needed. So the why I think is there.

The second question is, has the safety really been adequate with the new technology. They have really done I think an outstanding job of looking at the safety issues involved. Remember that they are trying to measure a safety component here that cannot be measured because it is so small and because there is no precedent for this. The kind of calculations that have done have really been built on standards in the industry and standards that the FDA has established in their guidance to industry, and they have done just an exceptional job of that.

I think what I hear from the Committee are a few things. One is let’s test the assumptions a little bit more around the kinds of procedures that are used. For instance, about the size of the DNA. They are just asking to test these assumptions a little bit further, and give us a little bit more comfort that the standards as established are going to hold up. I think that is important along the lines of a comment I made at the very beginning of this, and that is are we in some way setting a standard or setting a precedent for future evaluation of cell lines? And I don’t think we can get around that. To some extent what we do today will establish a standard, although it is going to change as it evolves.

The other issue I hear from the Committee members is trying to understand a little bit more about the technology behind this. What actually happens when you do the filtration and how do you add safety factors as opposed to multiply them -- again these grow out of standards that the FDA has set forth, but I think that the Committee needs to hear and has heard a little bit more explanation about what those are.

The key issue is that we are trying to establish is there enough information to allow initiation of a Phase I trial? I don’ think we should lose sight of that fact. This is really the beginning and the safety evaluation of this method of growing vaccines is going to be an ongoing and evolving event over quiet a period of time. The start has been really quite exceptional by both the FDA and by MedImmune and it is comforting to see that they come up with similar numbers for safety factors. In fact, the FDA numbers provide a greater safety factor than the industry, than MedImmune has. It has been a little more generous in providing those safety factor numbers.

So I think that it is really in play I think, that they have done an outstanding job. One thing that the company -- and I am sure they have thought about this and will continue to talk about this -- communications, and actually I think it comes to all of us in the medical community, communications to the general public about this method of growing vaccine, manufacturing vaccine, because it is new and there will be some misunderstanding from the lay press, so in the future our communications need to be extremely clear about what is being done and how the safety deliberations have been performed.

DR. MODLIN: Dr. DeBold?

DR. DEBOLD: II would just like to build on what Seth said. I think communication of risk is going to be very, very important, and this goes back to the original remark I made this morning about working with patients and consumers as it relates to safety and risk and FDA’s and CBER’s mission. This is going to be a tough one, I think, to work with.

I am still having difficulty understanding the differences in tumorigenecity across the various experiments that were described by FDA and industry staff. I mean there are some enormous amounts of variation in performance of those cells and I know that the FDA staff said they were not working with the MedImmune cells. Nonetheless they all come from sort of the same source.

This is going to be tough for people to understand and I think it is going to require some very targeted work to communicate satisfactorily.

DR. MODLIN: Good points. Dr. Cook?

DR. COOK: I have a couple of specific questions and then a comment. I’ll do the questions one at a time. One is, I asked after the MedImmune presentation about the documentation that had been provided by CBER about this illness in animals challenged with high MDCK doses, and the response was that there were some things associated with housing and superficial abnormalities in the animals, but then Dr. Peden presented a slide that I guess comes from the MedImmune data that there was an unexpected weight loss in the nude mice. It seems to me that is not so different from having an unexpected systemic illness in the animals.

I would just like to follow up on that and try to understand better what that means. It sort of hooks into Dr. Coffin’s comment about the issues of tumorogenecity and cell parts and other things that are known getting across in vaccine preparation have been -- the calculations are staggering, of course -- but probably the safety level is pretty high. But the things that aren’t known are obviously difficult to regulate and are they are difficult to look for, but we can’t pass up opportunities to look at things like sick animals being challenged with cells by maybe unexpected agents that we don’t know about, and that haven’t been tested for.

I think it would be worth getting a response about the explanation for the unexpected weight loss?

DR. KEMBLE: We do have the weight loss data -- if we can get that slide up to address this specifically, the one you set the stage for a couple of things. Thee were two studies involved with the newborn when we did not see the weight loss in the newborns. What we hoped to convince you of is the weight loss was associated with an operator error that happened during the Thanksgiving holiday when the animals did not get fed. It was after the time that the tumors, the nodules had been evident and she can explain the rest of the details for you.

DR. RICHMOND: The response to that question -- there was no MDCK-related abnormalities clinically. What we determined, and we of course were curious about the weight loss as well, is that it happened across all dose groups including the positive controls, the HeLa-injected group, it happened at week seven, and it affected everybody equally. So it was determined that over the Thanksgiving weekend the animal keepers had neglected to feed the animals so there was a correlation to that time. However, I think that we are covered because we have data to show that the tumor size of the positive control animals, the HeLa-injected group, continued to growth throughout this period. There was no change in that. Then most of the animals in the other dose groups, in the MDCK-injected groups, the MDCK nodules had already regressed by week four. So really prior to the incident where the food was omitted from the cages.

DR. MODLIN: Dr. Cook, are you satisfied with that explanation?

DR. COOK: Yes, and I have another question. This is one that I just failed to ask before. That is, the protein that comes through into the vaccine preparations -- I didn’t quite get that right. Is sounded like it was, and this may be wrong, half of a microgram per dose? Or is that way high?

DR. KEMBLE: That is correct.

DR. COOK: And I don’t have a sense of what protein concentrations in vaccine doses mean, but the question is other than worries about tumorigenecity and other kinds of things, that sounds like a significant amount of protein. How does that compare with other things like VERO-prepared vaccine strains? Is there any way to measure what that is likely to do to a recipient if they are getting a half a microgram of I guess this is MDCK mostly protein or what do you think the source is?

DR. MODLIN: Norm, do you want to address that? Actually it would be nice to have an FDA response in terms of comparison to other vaccines -- say oral polio.

SPEAKER: Egg-grown vaccines would be nice, too.

DR. MODLIN: Or egg-grown would be fine.

DR. KEMBLE: I can comment relative to the egg-grown vaccines. Certainly that product license in 2003, which was called the frozen Flumist was essentially allotillic(?) fluid straight with a little vaccine in it, but huge amounts of cellular protein. Our current product goes now through a centrifugation step and it is significantly reduced but it is on the order of magnitude of what we are seeing with the residuals here. And for the other vaccines we don’t have direct experience with those.

DR. MODLIN: There must be some information from the inactivated vaccines -- well, no, because they go through an additional step that would denature the protein -- I’m sorry.

DR. COOK: The only other thing I would say is more of an overall comment. I think it is obvious to an outsider like me listening to these presentations the importance of the science on both sides. Obviously MedImmune has done a good job, the due diligence, in terms of a lot of testing and lots of data from their perspective. But the complementary what I would call unbiased evaluation by the FDA investigators helps put a lot of that in perspective and helps explain some of the things like the sub-cloning of the MDCK cells. And this goes to a higher level than probably any of us, but I think the country probably doesn’t appreciate the kind of money it takes to do that kind of work and the ability of scientists who really get their work done in an environment where they are expected to do a lot of regulation.

If I think about it from our academic perspective, we have a spectrum of people, some of whom see a lot of patients, some of whom do a lot of science. We don’t expect them all to do the same amount of patient care and maybe from the FDA perspective, what they may consider is maybe they have people like this who are doing very important scientific work and how much their regulatory load should be given their limited resources and limited staff, with 30 PIs and what sound like to me a very small budget -- I know that is not their problem. They would love to have more money, but I think it is a big issue.

DR. MODLIN: Dr. Gellin?

DR. GELLIN: I have dealt on that and a couple of other comments from my neighbor here. I will start with the last one. I think that for the people who are not inside this room and tracking this, this may seem like some arcane piece of science. But we are all here and we recognize how important it is to try to chart our way through this area of uncertainty by the company and the scientists at FDA, and how that is being handled. I think that is yet another communication challenge to fill in one that Dr. DeBold mentioned.

I think, John, you set the theme of this session when you said in this vast uncertainty. Now the question for the FDA about how -- I mean the language in the background document was interesting -- theoretical risk, potential risk, all those kinds of things, because we just don’t know and we see all the fancy math with exponents that you are not really sure how many zeros are in there, but you know there are a lot -- on one side of the decimal place or the other.

Hoe does this compare -- again it is a different set of products -- with the discussion about the risk of TSE? Because there are similar kinds of approaches where there are lots of exponents thrown around. Are we in the same general ballpark or is that a totally irrelevant question?

DR. KEMBLE: I hesitate to try to compare those two. I think we have probably more, at least from the vaccine perspective, I think we have more data here and that the whole approach that Dr. Peden and others have taken as far as coming up with these calculations were really based on quite a bit more data that was generated at the FDA.

So I really can’t compare the two. The numbers are probably, if I remember, are probably in maybe the same ballpark, but how we derive these numbers versus how numbs were derived for what we did years ago when we presented the issue of TSE, I think that was more theoretical. WE have data that is in some of these calculations.

DR. GILLEN: I think that is probably the best answer we could hear. As theory is filled in by real data it helps you refine where you are. So that is helpful.

I have a question for MedImmune. It may just be remedial for me, but in your cartoon of how you picked your clone, could you do the same thing again and pick the clone? Because what I learned was Dr. Cook’s take home comment is that, at least in this cell line, they are going to play out differently. So is that lucky? Or could you do it again?

DR. KEMBLE: We don’t have an answer. What we can look back to is the productivity analyses where we looked at multiple clones and we saw differences. We would expect, if we looked at 2,500 clones for tumorigenecity, we would see some range. But I don’t know how that would all follow the high productivity clones at this point. We don’t have any data to support that one way or the other.

DR. GELLIN: I will just throw in that if you just look at the raw data that Dr. Peden presented, they had a 70 percent chance of getting the cell line they got. So it is not so surprising.

DR. MODLIN: Good point. Dr. Romero.

DR. ROMERO: I just have a comment. I think all of the questions have been asked. I think it is important not to lose sight of the route of administration of this vaccine, a point you brought up and many of us considered. This is not a parenterally-administered vaccine. This is a topically-administered vaccine, topically to the nasal mucos. So that needs to be kept in mind.

That being said, I agree and was concerned about the size issues of the DNA, which have been beaten to death and I am not going to go over that again. When I read the preliminary data or the preparatory data I think that we should have the technology, I hope we have the technology to shear those fragments down further. And I think the comments that were made about histone and the linear DNA that we are digesting in experimental models is not the same as what we are doing in this vaccine.

That being said, I have no other comments.

DR. MODLIN: Thank you. Melinda?

DR. WHARTON: This has been very interesting to learn what we know and what we don’t know about this very important area that we actually don’t know as much about as we would like to. I guess that is one of the things that has impressed me going through this.

I do think it is important that we keep in mind that at some point we may have an influenza vaccine that won’t grow in eggs. Keeping that in mind it seems like we do need to find a pathway forward on this one. In that light I appreciate all the great work that both the manufacturer and FDA have done to try to shed light and help us understand what is really going on here.

The two specific questions I would have had, had they not been addressed by the other side of the table, had to do with the size of fragments and whether or not a positive control would help us be more sure about what those lysate studies meant. The thing is seems like we need -- maybe need is not the right word -- maybe it is want.

The thing I want to understand better probably isn’t a necessary precondition to proceeding with Phase I studies. But it does seem that if we feel like MDCK cells are an important part of our national repertoire for humans with influenza, these questions that were outlined in the FDA presentation about what are we to make of all this - it seems that it is important that the research continue, because if in fact what one finds depends on which clone one selects, that is a little disconcerting.

I would feel better if I felt like the people who know the most about this could say they understood it. It is okay if I don’t, but I would like there to be more certainty in this area.

DR. MOUDLIN: Thanks, Melinda. Frank.

DR. DESTEFANO: My comments on my thinking in this right now. For tumorigenecity - it seems like for there is scientific data with the clones and such, is most perplexing now. The data that I have seen would suggest that it is very, very unlikely that a cell is going to make into adults. If that is true then perhaps, of the discussion about tumorigenecity and the clone may not be that big a factor to enter into this.

The other issue about oncogencity and DNA fragment sizes and such, that seemed to have generated the most discussion among the Committee members. Although from the FDA presentation, that was the one point that they felt the evidence as has been presented is most reassuring.

DR. MODLIN: Fair enough.

DR. KEITEL: Much of my questions have been answered or articulated repeatedly. I have just two follow up comments. The first is really more directed at the FDA and that is the threshold of 10 nanograms of nucleic acid was established somewhat arbitrarily for the inactivated vaccines. I am just curious as to whether a new stand is going to be - whether this will be precedent setting - a new standard will be established or determined for live virus vaccines based on the one nanogram or less of nucleic acid in this product - that is a precedent?

The second gets at risk communication. I am putting my hat on as a vaccine evaluator and thinking about how I would write the consent form this phase I trial. It should be a very interesting process to think through how confidently we are going through the minus 23 and the minus 15’s and unknown risk assessment, when the feeling remains that this is an issue.

DR. MODLIN: Good point. Excellent.

Dr. McInnes.

DR. MC INNES: Echoing everything that has been said before, and I think in particular I sat here thinking about how we could write a consent form - convey to people we don’t think we are reasonably confident. We don’t know.

We still are dealing with some of these issues and I think that is not to take away at all a very robust and beautiful of work that was put before us from MedImmune.

I do think we have an obligation to try to answer the potentially answerable questions. I’m not necessarily comforted by a division between Phase I and beyond studies if there are things that can be answered because I think the communication piece for the Phase I really can be very, very challenging.

I have a question about - there are clearly some things that could be experimentally addressed and those were put on the table. I have a question about why not sequence the cell line?

DR. KEMBLE: Would you like me to address that question.

DR. MODLIN: I’m sorry, Dr. Hughes, go ahead.

DR. HUGHES: Let me make a couple of comments before the gentlemen from MedImmune. If you assume that your error rate in sequencing is 10 to the minus 4, which is pretty good, you will have 10 to the 5 errors in the genome when you sequence them approximately.

The other question of course is what you use as your standard to compare it to. There is a draft copy of the dog genome but I don’t think it is from a Cocker and I think it is only a draft. Certainly one could look at the well studied and well identified oncogenes and there might be some virtue in that. But I don’t know how confident that would make me particularly if I was looking at something that I was reasonably confident had 10 to the 5 mistakes in it.

DR. MODLIN: Dr. Kemble.

DR. KEMBLE: That is actually a beautiful way to address the question which is there are technologies that are developing; deep sea glimpsing, digital subtraction of RNA’s, microarray analyses for pathogens, all of these are emerging and as they become to a level where we feel that they can be applied in this setting.

As I said in the presentation, I am not trying to be trite, safety is a primary concern for us and our products. We certainly will apply them and we don’t currently as of today, see the right answer that would help us assure that safety to that mix level.

DR. MODLIN: Thank you. Pamela, did you have anything else?

DR. MCINNES: I am thinking.

DR. MODLIN: You are thinking. While you are thinking, I might just add my two cents. I haven’t heard anything from any of the panelists that I would disagree with. I think this has been a superb discussion. I would like to congratulate the FDA for inviting the right people to the table today because I think the discussion has been terrific.

What we haven’t spent much time on is the public health need here. We have certainly talked around it. It is certainly part of the introduction here, but I would like to point out that there are alternatives, both in terms of inactivated vaccines, some of which are now growing in cell culture, as well as your continuing to produce the live vaccine in eggs. Although as Melinda points out, at some point in time, if we are dealing with a highly pathogenic avian strain the latter alternative may not be available to us. So it remains a very important issue from that standpoint, and I think perhaps the most important when we are talking about a compelling public health need for this technology for a live vaccine.

I think safety issues remain, although personally I don’t think I have heard anything today that would make me think that, although we have these incremental issues that we have identified that would give us some added level of confidence, we still have a large degree of uncertainty. I guess just based on the information I have heard today I think it is unlikely that any time soon we will identify any new information that would make us less confident in the cell line. I am just thinking for myself.

Are there other questions or comments? Is the Committee prepared to vote on question one?

DR. KEITEL: This may have been in the briefing materials, I am not sure. I saw the estimates of the confidence of having removed cells, but is there an end of process assay that ascertains the presence or absence of cells in the material?

DR. KEMBLE: We do not have one in place because it would be microscopic at this point in time. I can tell you we have work that it is not there but it is not a current release assay because it is not a particularly robust way of measuring that.

DR. MODLIN: You can’t open a large number of vials and spin them down and see what you see under the microscope. Dr. DeBold?

DR. DEBOLD: I think we have all figured out now that all sub-clones are not equal by virtue of the FDA’s presentation. So is that question up there specific to the sub-clones developed by MedImmune? If so, is there some way to make sure that that is the sub-clone that we are voting on?

DR. MODLIN: I think the answer would be yes. Any further questions or comments? If not I think we will proceed to a vote. We will take the vote all at once and ask each of the voting members of the Committee -- help me, Christine, is everyone permitted to vote except for Dr. Hetherington?

Okay, those who would vote yes on question one -- are the preclinical data discussed today adequate to support the initiation of Phase I clinical studies -- I assume that means soon -- for live attenuated influenza vaccine in NBKC cells? Those who would vote yes, raise your hand.

(There was a show of hands)

Okay, Dr. DeStefano, Dr. Wharton, Dr. Romero, Dr. Gellin, Dr. Cook, Dr. Modlin, Dr. Hetherington, Dr. Coffin, Dr. Sanchez and Dr. Self.

Those who vote no?

(There was a show of hands.)

Dr. Hughes, Dr. McInnes and Dr. Keitel.

I think that is everybody who voted.

Thank you.

Let’s move on to question two. I think that most of these issues have come out in the discussion are quite relevant, regardless of the vote we just took. I think the answers were pretty clear, or the discussion was pretty clear. Did you want to raise anything else that might help focus the Committee in terms of this second question?

DR. BAYLOR: I think we really have to look at timing. If we are talking about doing a Phase I study as expressed in that first question. We know if we are moving down this road, eventually larger studies are going to be necessary, and then potentially licensure of this vaccine. I have heard the discussion prior to the vote on the Phase I study. Are there additional items that you really believe should be done prior to it, because even some of the comments were not really timed. Some of them were really into the future. It would be helpful to understand sort of the timing of those studies proceeding past Phase I, if there are things that are critical that you can recommend are accomplished prior to moving into those studies.

DR. MODLIN: Why don’t we go around the other direction then. Maybe start with Pam.

DR. MC INNES: It is the question on protection of longer DNA pieces and the lysate studies I think should be done, but I think they should be done just before Phase I. That is my caveat.

DR. KEITEL: Not really understanding the difficulty it seems there is a lot they can determine in that larger distribution the fragments of remaining DNA. That would seem to be a reasonable request based on the concern expressed by most of the Committee members.

DR. MODLIN: Are there any other studies that would be reasonable to request prior to the Phase I study. Dr. Cook?

DR. COOK: I personally can’t remember all of the comments that have been made, but I think it would be reasonable for the FDA to go back through the transcript and find out all of the concerns that people have raised. It seems to me that a Phase I study is meant to find out if the vaccine is or is not immunogenic in people if they are willing to take it. But I don’t see any reason it should progress beyond that until they have done a thorough review of those questions and gotten to the point where they get a hundred percent consensus of whatever committee they ask that they ought to proceed beyond Phase I. I personally don’t see what’s the rush until those questions are answered. It is not unreasonable to find out if it induces an immune response in humans.

DR. MODLIN: The only thing I might say is, of course the focus of the Phase I study is usually safety, almost always and the numbers that might be involved may or may not be adequate to show immunogenesis.

DR. COOK: I would say in response to that that it is probably, at least from my personal perspective, that is sort of a backwards way to think about this vaccine, because you can’t address safety on an uncommon event in a small number of people. That is not going to work. So there is going to have to be adequate comfort with the safety before proceeding. I would think of asking whether the vaccine has any chance of working would be the first question. Then the safety is going to be all these calculations we have been talking about, what is the likelihood that something will happen in a hundred million people? You are not going to do a Phase I study in anything other than a few.

DR. COFFIN: And ten years from now -- there is no conceivable way this could be evaluated in a Phase I study.

DR. MODLIN: You really can’t design a safety study, can you?

DR. BAYLOR: I could make a comment and push a little further. With a Phase I study we are not going to be able to address the type of safety issues that we are talking about. Basically it is a proof of concept, and that is why the firm wants to do the study, to make sure that this vaccine, grown in MDCK cells, will behave just as the current vaccine grown in eggs. They have the animal data to demonstrate that that is true. But moving further, and that is the challenge for us, if we move forward in this study how do we get to the next step?

Are we comfortable getting to the next step of doing those larger studies and then eventually licensing this vaccine, and even with licensure we know we cannot do a study large enough to address these issues. But we have to -- as I believe Dr. Cook indicated -- with the calculations we have made -- do you believe that we are at least on the right track the way we made calculations and the assumptions we have made. I think that helps us and that may help us to move forward.

DR. COOK: Again, from my perspective, there are two kinds of safety. One is the administering of vaccine -- does it elicit some kind of allergic response or some kind of immediate Phase I typical sort of thing. That is different from all these other things we are talking about. I think those are going to end up being more a comfort level thing for acceptance of people taking the vaccine for trying to market it in some way that makes sense.

So everybody has kind of gotten onboard with the idea that you have done everything that you can realistically do. You have addressed all these concerns and now you can comfortably go ahead and try to convince people to take it. We are not doing so well even with flu vaccine as it is. We are lucky to get 50, 60, 70 percent of people even who are in the health care professions taking the vaccine.

So if there are these kinds of questions, and everybody can’t come forward and say we are completely comfortable with this, it is going to be very difficult.

DR. SELF: Just piling back to question number two, I guess I would put on the table that the one thing that concerns me is the longer term risks for some of these effects. To the extent that, before moving into larger studies, some animal studies with much longer term follow-up to try and look to see if there are any long-term effects, I think that would be the one category of study that I would place in number two.

DR. BAYLOR: If I may, Don, push you a little further on your issue also of the public health. The bottom line is benefit risk here. If we go back to that premise, what are our options? Do we have any options as we try to evaluate the benefit/risk of going forward with this knowing that we haven’t resolved all of the theoretical risks and probably will not. But the benefit of going forward -- I think that has been outlined pretty clear from the sponsor as well as us.

DR. MODLIN: I was thinking the same thing and I was a little hesitant to raise it. Ultimately it will determine what is in the label, what the vaccine is labeled for, and I think we would all agree -- as was pointed out -- that the risks/benefit ratio changes dramatically in the presence say of an avian influenza outbreak or a pandemic influenza. I guess that issue -- you would have to do adequate testing of the vaccine ahead of time in order to even label a vaccine for that purpose if the label were to be restricted to that purpose. WE have licensed an avian vaccine already. To me, a vaccine frown in cell culture would have a distinct advantage over the currently licensed vaccine. But we are taking things from a different direction there. I think we all recognize that is an option that needs to be put on the table.

DR. HUGHES: Obviously we are all deeply concerned with risk/benefit. I think maybe in the issue of the Phase I trial is subsumed the idea of whether or not we, as a group, could feel comfortable going forward with a large-scale trial without some of the additional analysis that a number of the Committee members include. So I think the real question is whether you do those before you do Phase I or after you do Phase I. I would be surprised if the Committee would want to go forward with a large-scale test without that information.

So I guess I would say to you, I think it unlikely that even a modest sized Phase I trial is going to tell you very much about safety here. I think the reasoning behind that is clear. That is the primary reason for a Phase I trial. This is not a new product. This is an old product made in a new way. So I think it is very likely that it is going to be effective. If not the flu part is not going to substantially different from the flu part that is made in eggs. It is really a question of whether or not the stuff that comes along out of eggs and how easy it is to manufacture and how fast you can make it.

The real question, I think, in terms of the question that is posed there is whether or not by saying we would like data before we go to Phase I as opposed to before we go to a large-scale test, is actually going to change the time line very much. I guess I am not convinced that it is. If it isn’t going to change the time line I would personally prefer to see us choose the safest course, which is don’t put it into humans until we have what would seem to me to be relatively easy to obtain additional data. It is clear that the MedImmune team has a lot of technical power. I would expect in a matter of months that they could answer our questions and answer them clearly.

If that expectation is reasonable I don’t think we lose much in the temporal sense by saying before we go into humans we know all the answers we can easily ask. It is really on that basis -- if you are asking me whether I expect anybody to get hurt in a Phase I trial of this sort, the answer is no. But the question is really what is the benefit of doing the trial before you have the additional answers? Right now I am not convinced there is much, because I am not convinced what s going to accelerate the time to approval of a large-scale study, which is what obviously we all care about.

DR. HETHERINGTON: I think the answer to that is again the risk/benefit ratio, the smaller number of patients you are exposing means that your risk is low. And I don’t think we are talking about -- although we would like them to do additional studies on the size of the fractionated DNA, I don’t think we are talking about moving the safety factor by more than a few orders of magnitude at most. And even that doesn’t register anything on the risk scale to the small number of patients who would be participating in the Phase I trial. I assume what the company is trying to do is at least show that the immunogenicity that they can detect in a Phase I study and the lack of a common severe enough side effects that might be local or systemic in nature, but clearly is not the kind of risk we are talking about today, does not outweigh the kind of immunogenicity data that they would get, and they would therefore have a good rationale for moving forward to the next larger study.

I think from a vaccine development perspective, while you might say their timing might not change, it does not make a whole lot of sense to hold them up on a Phase I study which is clearly a go-no go to the next big study, at the same time they could be going on and answering these other questions that are important to the committee. So if you look at it from the risk/benefit perspective and the difference in risk that you might be applying to a small group of patients, with or without that information. I don’t think it moves the needle very much. So I don’t see the impediment to going forward with a Phase I.

DR. MODLIN: I think what Seth is doing, Dr. Self, is pushing back a little bit here in the sense that -- let’s make the assumption that we have asked the company to do these studies and the results are maybe unfavorable, that you see larger numbers of fragments or a different distribution of DNA fragments in than you expect to see. Would that change your mind with respect to making a recommendation about a Phase I trial?

DR. HUGHES: My feeling is actually I would expect that if the distribution of fragments would be unfavorable that the company might want to change its manufacturing practices in a way that that would change. I would feel more comfortable giving that opportunity before we begin the human tests, myself. And others should comment.

The positive controlled experiment might take months. The DNA size experiments should be trivial -- days -- with the technology that is now available. I think there is a very good chance they are going to tell me something that will make me feel good. But I think if they tell me something that would make me feel bad then I would expect and hope that they would feel bad, too, and they might want to change their benzonase treatment or add something to the benzonase.

This is not -- I am sure that the company would very much like this to be the final form of the product. But I think we are in a position now, or they are in a position now to adjust and make minor adjustments in this in a way that might benefit them. I must say I actually see more upside in this than downside, because I think they are going to get a good answer if they do this, and then we are all going to feel better. It is the unknown that makes us nervous.

DR. COFFIN: I think we are in general agreement that there really is no perceptible risk from these issues we have been discussing in a Phase I trial. I think it should be more a company decision as to whether they want to take the risk -- because they are certainly going to be asked to do these studies before they go into anything larger. Whether they want to take the chance of finding something there that will cause them to have to go back to restart as compared to being able to proceed if all of that works out kosher in the way we expect. But I think the decision in a sense as to whether it takes that operational risk is more up to the company than to us.

DR. MODLIN: Good point. Other comments? Norm or Dr. Weir, did you want to try again?

DR. GELLIN: Let me give Norm some time to think. I want ot put together two comments, one that Pam had and one from Norman. I think they are relevant and important. It has to do with the sequencing. I think the comment that we should answer everything that is answerable is an important one. Exactly when that has to happen I think is what we are talking about now as far as the sequencing of the studies. But I think if there are questions that are out there, and that may be what you asked for, to sweep through the transcript and everyone has been taking notes, that sooner or later those are going to have to be answered, because the question from either the potential volunteer or the potential patient will be have you done everything you could to make this as safe as possible? So I think we have that obligation.

But Norma’s response to my CSE question is along the same lines. The more there is data in there, the less these numbers become just back of the envelope calculations. So I think it is putting those things together that are going to help show all of us that we have done all that we could.

Now, Norman, you have had enough time to think.

DR. BAYLOR: Think so?

DR. SELF: I guess I just want to inject here, we have gotten kind of hung up, and perhaps I have, too, on the distribution of the size of the fragments and all of that. But there is a fair amount of animal model data across three species and that was really pretty squeaky clean. So if you think about what the standards are from moving to small Phase I studies, my vote is informed probably more by that than more arcane discussions about DNA fragment size and what that really means. I want to know that, but I am not thinking that that should be our main thing.

DR. MODLIN: That is a remarkable comment from a mathematician. DR. COFFIN: I would like to say something about the sequence -- I think going forward I think this will be very important thing to do, but I don’t think it is anything we can do anything with really right now. IN that sense I am in agreement with Steve. But the thing that you cannot ask right now a specific question that will be answered by it, but it might well be that having a sequence will allow you to frame certain hypotheses which you can then go forward to test. And I think that is not a standard we can apply to this at this particular time.

I predict that five years from now that is a standard we would be able to apply, but I don’t think we can do it now. And I think for cell lines like this, which will become widely used, I think this will be a very important thing to do. It will be very important to compare tumorigenic and non-tumorigenic clones and really try to come both an understanding of what is happening and also almost certainly for new ways to devise incisive tests to perhaps identify what the relevant oncogenes really are in a cell line and ask whether that DNA is getting in or whether you can use cells that don’t express this particular mutant, if it is the mutant or this particular transcriptional variant. But this is all stuff that I raise for the future.

I don’t think the issue is now. The only thing that I think sequencing might do for you is to identify adventitious agents that you otherwise don’t see. Again, it is a fishing expedition at this stage. It doesn’t make real sense to send MedImmune down when there is no clear answer at the end. That is my perception.

DR. MCINNES: I think it may make a lot of sense in the context of a massive federal investment into cell culture-based vaccines. WE happen to have this example in front of us. I am not suggesting that we have to have all guns focused on this particular issue, but I think it may be extremely important to turn the heat up on the stove on developing some of these data that are going to be critical for us.

DR. COFFIN: I completely agree with that. I just don’t think we can use it as a standard right now, for this specific question. To push this forward as fast as possible for FDA and for industry I think is something that really should be done as quickly as possible.

DR. MODLIN: Norm, would you like to have a final word?

DR. BAYLOR: I guess we have taken this as far as we can. I think the vote fell where it fell, but I think the discussion after the vote was most important and I think we will be able to make some decisions once we analyze those comments that were made. But those were really good comments after the vote and we will take them into consideration as we move forward.

DR. MODLIN: Excellent. Anyone else have anything? If not I would really like to congratulate both the sponsor and the FDA, and particularly the panel members. This really has been a superb discussion. Thank you all and as far as I am concerned we are adjourned.

(Whereupon, at 2:40 p.m., the meeting was adjourned.)