FDA Workshop
Anthrax Vaccines: Bridging Correlates Of Protection In Animals To Immunogenicity In Humans
DEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR BIOLOGICS EVALUATION AND RESEARCH
Grand Ballroom
Hilton Washington North
620 Perry Parkway
Gaithersburg, MD.
Printable Version (PDF, 255 KB)
Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®.
Friday, November 9, 2007
- Session 5: Animal Models for Post-Exposure Prophylaxis
- NIAID Perspective on Rabbit Post-Exposure Prophylaxis Model Development, Judy Hewitt, Office of Biodefense Research Affairs, Division of Microbiology and Infectious Diseases, NIAID
- Development of a Rabbit Post-Exposure Prophylaxis Model for Testing Combinations of Antibiotics and Vaccines against Inhalational Anthrax, Jason Mott, Battelle Biomedical Research Center
- Session 6: Panel Discussion on Post-exposure Prophylaxis, Moderator: Ed Nuzum, NIAID
- Post-Exposure Prophylaxis: Issues to Consider when Developing a Correlate of Protection for the PEP Model, Bruce Meade, Meade Biologics, LLC
- Session 7: Meeting Summary, Moderator: Ed Nuzum, NIAID
- Summary comments from panelists
P R O C E E D I N G S
8:34 a.m.
DR. MEADE: Good morning. It's about 8:30. Can we go ahead and get started? Good morning. My name is Bruce Meade. Today I am serving in the function as a consultant to DMID and will help moderate the morning session on post-exposure prophylaxis. Before we get started Freyja asked me to make three announcements and hopefully I will get them correct and, if not, she'll make sure I get them right.
First is I understand there are some extra copies of the meeting materials, the handouts, the slides that you all received. I think there are a few extra copies that will be out in the lobby. Feel free to take one extra copy for a colleague if you would like one.
I understand, again as you heard, that the transcripts will be available, my understanding is, sometime within the next month or so. The instructions on how to obtain those will be available on the FDA website. I guess you can start checking that in a few weeks for that information.
I understand those of you who need taxis should make arrangements directly with the hotel staff. Individuals meeting are really feared up to do that so work directly with the hotel staff.
Again, I think just to briefly introduce this morning's sessions we are going to change direction a bit and deal with another important aspect of the anthrax and that is on the post-exposure models and approach and approaches to dealing with vaccination after event if a possible event should occur.
There will be three talks. Judy Hewitt, Jason Mott from Battelle, and then I will finish up with mostly asking you some questions at that point. The first talk will be Judy Hewitt from DMID who will talk about the development of the rabbit pet model.
DR. HEWITT: Thanks, Bruce. Good morning, everyone. What I want to do today is tell you a little bit of the story about how we developed this model. I'm not going to get into any data. I'm leaving that to Jason.
First I would like to start off with some of the assumptions that we made going into this program. In the beginning we anticipated that we would need two models for licensure, the rabbit and the nonhuman primate. That was really based on the 2002 workshop and discussions.
Back in 2003 when we started this effort the post-exposure prophylaxis, or PEP indication, was in our minds going to be the second indication that would come after licensure of the vaccine for the GUP indication.
Also, at that time we anticipated developing a nonhuman primate model ahead of a rabbit model and that was based on some preliminary studies that have been done by Friedlander in the early '90s predominately aimed at licensure of various antibiotics for anthrax, but there was also a component of those studies that had some vaccine added to the antibiotic regimen.
Because the antibiotic Friedlander used at that time was cipro we assumed that would be the antibiotic that we would use for our model development efforts.
Of course, in science there are always changes to your plan so PEP became the highest priority around 2003 when the bioshield legislation was first proposed and it was clear that there was an intention to stockpile the vaccine and it would be used basically for emergency use or under emergency use authorization.
The push was to get the PEP model developed first rather than having that PEP be a follow-on indication. So we had a proposal of how we were going to go about developing the PEP model and with the importance and urgency of developing this model we deleted one study that had originally been in our plan and that was to determine the start time of the antibiotics and the vaccine regimen.
Instead of performing a study to actually decide what was the most appropriate time to start antibiotics and vaccination, we just discussed it, picked a time, and went with it.
Of course, the antibiotic changed. I should back up and say we started our efforts in parallel with both rabbits and nonhuman primates. You are going to hear about the rabbit model today because that has essentially been completed. We are still working on the nonhuman primate model.
In the rabbit model we changed the antibiotic to levo. That was based on the rabbit's tolerance of the antibiotic. Then, of course, as is typical in science, we got some unanticipated results.
Basically did a study where we got some very promising results but it fell short of statistical significance. That required a little bit of refinement and repeating a study so that added a little bit of time to our plan. At any rate, things moved on.
The other point I would like to emphasize here is that the development of this model the goal was really to be able to demonstrate added value of vaccination. It was not strictly to use the human regimen in terms of the antibiotic and the vaccine.
In fact, I think until the human regimen or human dose of the vaccine is really determined or humanized dose in rabbits if you will, these studies will have to be repeated. At any rate, we feel that the model is very well developed and ready for final drug product to be tested at the appropriate doses.
This just gives you sort of an overall summary of the sort of level of effort that went into this project. I have some proposed and actual statistics here. The duration of the end-life studies we anticipated to do these in very rapid order but it actually took us much longer.
My second line here, the duration with the reports, is really important because it is one thing to do the studies and gather the data but then you also have to get the reports into FDA so that they can look at the data in a very rigorous manner. It has been a real considerable effort to get all these reports completed and into FDA.
Originally the original plan was seven studies. We dropped the one right away, as I mentioned, but because of the unanticipated results we actually ended up doing nine studies in the end. The number of rabbits you can see doubled basically. I think this is kind of an important point.
It gives us a huge amount of confidence in this model in that quite a few studies have been done. They have behaved very well from study to study. There is actually a large number of animals in these studies.
Of course, originally we were only going to test two vaccines and we ended up testing three. As I said, they behave very similarly and Jason will show you that data.
I want to introduce this concept of TRLs or technology readiness levels. It's something that the Government has used in the past to assess the readiness of particular countermeasures or, you know, in the military things like tanks. Are they ready to be bought and purchased by the Government.
In an effort to support planning for bioshield stockpile procurements people are always asking the question, "Do we have the animal models that are going to be needed to get these products licensed?"
We developed a series of levels here and this is just sort of a very high-level summary of what the animal model TRLs represent with one being your very most basic studies and 9 being GLP studies that are going to be sufficient for pivotal studies for licensure and all the various steps along the way always building on your model, getting closer and closer to your end target.
So a couple of conclusions then about our model development program here. I gave you a sense of sort of investment and effort spent on this program was.
We have basically progressed this rabbit PEP model from something like a TRL 2 or 3 all the way up to an 8 or 9 depending on whether -- I mean, in my opinion the model actually is sufficient to support pivotal studies and it is a matter of the product being at that level as well, final drug product consistency lots.
All of the studies that we did except the very first one were GLP and the result is a very well-characterized model. All of these reports have been filed. We have a master file with FDA so all of the animal model development work that we've done is filed with FDA through our master file.
The studies that have single vaccine efficacy testing in them have been filed through the various sponsor INDs. Of course then we grant the IND holders letters of cross reference to reference our master file for the information that they require to support their pivotal studies in the end.
I have to end with this slide which is a picture of the nine final study reports. This is all very neat and tidy. These are all completed. They are actually all now in the hands of FDA as of this week.
The fun is going to begin in terms of looking at this model and the data and assessing whether products are ready to be -- whether studies are ready to be considered pivotal to support the PEP indication.
With that, I am finished and I'll take any questions.
DR. NASS: I am unaware of strong evidence that treatment is needed in the time frame that a vaccine is going to become effective post-exposure. I know that was our assumption at the time the letters were sent but of the 30,000 people who may have been exposed who did not take long-term antibiotics and did not take vaccination, none of them became ill. So what is the data that shows that we actually need post-exposure vaccination?
DR. HEWITT: Well, I'm not here to address what the need is. I'm here basically to tell you that we have a model that can support that. In terms of stockpiling a vaccine the intention is that it would be used only in a post-event scenario. The need is an unknown at this point.
DR. NASS: Well, in order to get something licensed you have to have an indication so I would think it would be a critical question for you to answer.
DR. HEWITT: Right. We have a model that can support -- one model that can help support licensure for this indication.
DR. NASS: You don't have an indication.
DR. HEWITT: I appreciate your comment.
DR. MEADE: Our next speaker is Jason Mott from the Battelle Biomedical Research Center in Columbus. He is going to talk to us about all of the very difficult and important work that they have done with this post-exposure model.
DR. MOTT: First of all, I would like to thank Judy and Ed and Freyja for the opportunity to be here and everybody else for coming to see this talk. As Bruce mentioned, I will be discussing a lot of data that was derived over the last couple of years on the development of a post-exposure model in the rabbit.
As Judy mentioned, the post-exposure is one scenario that we were looking at that has been placed in a high priority just in case we do have another accidental or deliberate release of Bacillus anthracis spores to the public.
The whole premise of our studies were based on a combination therapy to look at the ability of the antibiotics to kill the spores that germinate after an exposure as well as a vaccine to develop a protective immune response to take care of the latent spores that were left in the lungs.
The current antibiotic regimen is 60 days. As we have heard, compliance is often an issue so certainly the added benefit of the vaccine will help those who do not take a full 60 days of antibiotics. They are reported to be less than 50 percent compliance following those last attacks in 2001.
The rabbit model, as Judy mentioned before, was worked through over a period of about four years with countless hours and discussions with an animal working group to help us show added benefit of the combined therapies with success being defined as increased survival in the rabbits following a potential lethal challenge.
Reproducibility was obviously critical for us and we were looking to establish a model that could be applicable to new vaccines or therapeutics. A lot of the initial goals in the original proposal were developed and individual studies were designed to help us evaluate each one of those goals.
As Judy mentioned, one of the goals which was determining optimum antibiotic start time was not performed. However, we still need to look at optimum antibiotic dose to allow us to get partial survival or death in rabbits post-challenge and post-treatment.
The studies I'm going to go through will help to address each one of these goals that you see here. The knowledge of these first set of goals was then used to do a combined study with antibiotics and vaccine with an aerosol challenge in the rabbits.
I have a lot of data to go through in a short period of time so if I don't cover something and you have questions, just find me or e-mail me at some point to help clarify.
The first study was a toxicology tolerance testing and pharmacokinetics study in the rabbits using two doses of antibiotic. It's a nonchallenged study. We use levofloxacin in these studies. As Judy mentioned, cipro was dropped out due to tolerance issues.
The study design is shown in the chart where we had two separate antibiotic regimens that were modeled to give us plasma levels that would be similar to what we would expect in humans.
There was a split regimen of 40 and 10 and then another split regimen of 20 and five mg/kg given about 12 hours across 14 days. We did take blood for plasma antibody levels. As you can see it bears time points as well as looking at CBC and clinical chemistry for the toxicity issues.
What we did find was that with levofloxacin there were no toxicity issues. We did not see any clinical signs during treatment. No lethargy, hematuria, respiratory distress or weight loss. We also did not see any significant changes in the CBC or the clinical chemistry data. Then on gross necropsy and histopathology there were also no signs attributable to the administration of the antibiotic.
The pharmacokinetic data did show that the Cmax or the P concentrations were found at about a half an hour to three hours post antibiotic treatment. At the high dose regimen we did see that the Cmax was at least 10-fold higher than the reported MIC50 of .12 micrograms per mil.
We also determined an AUC/MIC ratio at these two doses and found that at the high dose there was a ratio of greater than 125 which indicates a very high therapeutic potential. We also chose that dose for subsequent studies because we would expect that would be a dose that may interfere with vaccination if that is going to occur.
With that said this chart shows a little bit of the actual data from the two groups in comparison to some data that is in the literature for humans. We can see at the levofloxacin split regimen of 40 and 10 we have a Cmax of about 3.21 and AUC of 15.8 and a half-life of 2.6 hours.
With the dose that was half of that we see a similar half-life. However, we see a Cmax and AUC of about half that like we would expect. In comparison to the humans we don't see an exact plasma level that we would see in humans but, as Judy mentioned before, that really wasn't the complete goal here. It was just to try to make sure that we maintain antibiotic levels throughout the studies at levels that were above the MIC50 or MIC90.
Then we moved on to a second study which also was a nonchallenged study. It was a concurrent antibiotic therapy and vaccination study that also looked at various vaccine regimens and doses.
The objective here obviously was to assess the tolerance and toxicity of the dual treatment with antibiotics and immunization as well as to evaluate immune responses in the various regimens in order to determine a rapid response that we could expect to be utilized in subsequent challenge studies.
We also looked at the peak and trough flows of antibiotics. Once again, we did look at hematology and clinical chemistry data on these animals. This is a breakdown of what the study looked like and you can see that we had nine groups of animals with up to 24 animals in a group with various vaccination days all the way to zero and 21 and various vaccine doses of 10 or 50 micrograms of RPA.
We also had in group 9 one group that had the high dose of antibiotics added for 14 days with zero, seven, and 10 microgram vaccine dose. We also took blood at various time points, as you would expect, for ELISA and TNA analysis and plasma antibiotic analysis as well. What we saw here was that dual administration of the antibiotics and the vaccination did not have an effect on each other.
The levels of the ELISAs and the TNAs that were seen in comparable groups as well as the antibiotic levels did not change from what we had expected from the previous study. We also found that two doses was better than one as we expected and that boosting at day zero and seven produced a higher response than zero and five.
The 50 microgram of RPA group given on days zero and seven did produce a higher level of ELISAs and TNAs. However, it wasn't significant when you compared that to a similar group receiving 10 micrograms. We did not have any tolerance for toxicity issues in the rabbits, again, and we did have peak levofloxacin levels similar to what we had in the first study.
This is just one of the figures that was put together for the study that further is evidence of the immune response. The 50 micrograms of RPA given on days zero and seven on the top in red and you can see, as I mentioned before, there was a slightly higher level of immune response for the ELISAs and the TNAs. This is specifically for the GMC and for the ELISA response.
The other two lines that we can see here are zero and seven regimen with 10 micrograms rather than 50. In one of these groups, this purple group here, was given antibiotics at the same time so we can see that there was no ill effects of the dual treatment on these animals for the immune response.
We then moved on to a post-challenge antibiotic duration study. The main objective here was to determine a length of antibiotic to be given to these animals post-treatment that would allow us to have no deaths during the treatment but a less-than-optimum survival or less than 100 percent survival once the antibiotics were removed.
We also did the normal assessment of immune assays hematology clinical chemistry, bacteremias as well. The study design here was to take New Zealand white rabbits challenged with an aerosol LD50 dose of approximately 200 LD50s on day zero and then administer antibiotics for five, seven, or nine days.
What you will see is a little different here that we used levofloxacin at 50 mg/kg once a day rather than a split dose. Judy alluded to this earlier that we did have some issues part way through so I'm concentrating on the study that we used to determine antibiotic dose for the more pivotal studies.
The animals were bled for plasma antibiotic level as well as ELISAs and TNAs. These animals were watched for 28 days post-antibiotic treatment. We also did gross necropsies found dead to help us determine the cause of death was anthrax and make sure of that.
Results showed that observations on the animals as far as clinical signs were absent until the antibiotics were stopped. Most commonly we saw lethargy and anorexia, normal clinical signs. We did not see anything on clinical chemistry data. We also saw minimal changes in the CBCs once the antibiotics were stopped.
The typical changes were increases in neutrophils as well as decreases in lymphocytes post-exposure. Pharmacokinetics were as expected. We did have levels similar to the previous studies with the levofloxacin. As far as survival the important part once the animals were on antibiotics and post-challenge we did not see any deaths until approximately several days post-antibiotic treatment.
When we look at this table we see the results from the five, seven, and nine-day treatment groups and we see that with the five and seven we had over 90 percent mortality. With nine days we had about 70 percent. What you will notice here, too, is that when we look at the time to death, the time to death in those groups was obviously delayed.
There were no deaths during the antibiotic treatment. You can see that compared to the control nontreated groups we got a time to death that we expected would be good to allow a protective immune response from a combined treatment study. You can see there was a large range in death in some of the groups with one of the animals dying out close to 30 days.
We used the data from these first three studies to help us design the final more pivotal study where we combined the antibiotics and the vaccine together post-exposure. The main goal here was to determine if the combined therapy allowed the animals to survive long enough post-antibiotic treatment for that vaccine to kick in to have an effective immune response.
We also did the normal battery of tests as far as plasma antibiotic levels, ELISAs, TNAs, clin chem, CBCs, temperatures. The study design here was to take New Zealand white rabbits again and challenge on day zero with 200 LD50 of spores.
The antibiotics were given, once again, once a day at about six to 12 hours post-challenge at 50 mg/kg for seven days. Then they were vaccinated. Here it says with "with what?" I'm not sure that was really supposed to be in there. What it really means is that we used one of three vaccines in this study. As we go on we'll see that we repeated the study three times.
We did the complete gross necropsies as well. This is the basic study design of these three studies with the vaccine. What we see here are five groups of animals treated. The first three groups were treated with a combination of antibiotics and vaccine, the 50 mg/kg antibiotic for the seven days.
Then for the vaccine we used what I am describing here as a high, medium, or low dose for the studies. Group four shows an antibiotic only. Group five are the untreated controls. The vaccinations were also started at six to 12 hours post-challenge at the same time as the first antibiotic dose.
What we have done, as I mentioned, we repeated the study no less than three times using three different vaccines with the 50 mg/kg levo dose. We have seen very good reproducibility. All nontreated control animals have died through these studies. That's 28 out of 28. You can see that the time to death is very short of 2.4 to 4.1 days.
Mortality was absent in all the treated groups until cessation of antibiotic just as the previous study that I had described with a range of 11 to 28 days post-challenge. The vaccine combination groups did show that they had a significant amount of protection in combination with the antibiotics or with the antibiotic only.
Approximately 54 percent of animals in the antibiotic only group died post-challenge where only 9 percent in the vaccine groups. We can see that there was a slight dose dependency with four, seven, or 15 percent in the high, medium, and low doses respectively.
This is a figure that shows the mortality rates of the three studies in one. As we expected, you can see there was a rapid death rate in those animals that were not treated with a 2.4 to 4.1 average, as I mentioned before.
There is a slight grouping here of the three antibiotic only groups from the three studies at about the 50 percent death rate. Then there is significant changes here of the survival in the combined groups.
Bacteremia was also looked at in these studies and we did see that a very high percentage of animals that died on the study post-challenge or post-treatment were bacteremic. All over it was 93 percent. What I do want to note is that while the animals were on antibiotics very few, one out of 162 animals, showed a positive bacteremia.
In addition, only six out of 173 survivors showed a positive bacteremia at any time point. The bacteremias were not only taken at the time of death but also at various time points throughout the studies.
Anorexia was the most common clinical sign and that was more prevalent prior to death and post-treatment if they were treated. Temperature changes were also noted. We took temperatures twice daily from little transponder chips that were implanted into the rabbits. However, if you saw a change in temperature of 2 to 2.5 degrees it was very predictive of mortality in those animals.
Generally animals that were on antibiotics and vaccines, either of those groups did not show any changes in temperature while they were on it but prior to death if they did die it did become prevalent.
A mean peak in trough levels of levofloxacin were routinely greater than the MIC90 at all time points that they were collected except for the first initial trough collection. As far as the immunogenicity the anti-PA and ELISA and TNA was detectable by 10 days post-vaccination in those treated groups.
As we expected, antibiotic only groups if they did survive did not show a development of a protective immune response. It was significantly different than those animals that received a combination of antibiotics and vaccine for the levels.
What we have done is take all this data to help us define the model which we hope will be used to design studies testing other vaccines or antibiotics as needed. Certainly for submission to other regulatory agencies such as the FDA we would have to do this similar study in this model that we feel is very well characterized at this point to use their final product formulation.
I do have a couple more slides showing some of the information. This is similar to the earlier survival or mortality graph that I showed with Kaplan-Meier survival curves. It shows step by step when animals died for each of the groups. It looks very similar to the earlier.
We do have a geometric mean curve for the ELISA data for groups one through three for the three studies. There does appear to be a slight dose escalation there. The one complicating factor for these studies versus other normal vaccine studies is that we did have the challenge so we have to throw that in there.
It makes it a lot harder to interpret the levels of the immune response as well. This looks very similar to the geometric mean curve for the ED50 for the TNA data for the same animals.
There were some trends that were seen during these studies that we have taken for development of other rabbit anthrax models. Bacteremia is certain, as I mentioned, a very good predictor of death in these animals with over 90 percent of animals that died showing a positive bacteremia at some point.
Temperature is also something that we feel very confident in as a predictor of death with animals that show a 2 to 2.5 degree increase or decrease in the body temperature just prior to death. It's very predictive. It usually happens within two to three days prior to the animal dying.
Then for hematology we do see increases in neutrophils, decreases in lymphocytes and then the CRP reacts to protein which is an acute phase reactant. Usually it's going to be something that increases its levels during inflammation and infection and that is another thing we've gone to.
Baseline levels of the CRP in these animals are normally less than detectable limits. However, as these animals get sick to the point of dying the levels constantly increase at the time of death.
This is just a slide to show that the proposed path really wasn't what happened over these last couple years. Originally we did have six studies that would be done and two of those at the end would be very similar.
As Judy mentioned, we had a small hitch in the road here that made us go back, reevaluate what we were doing from the previous study, run another study until we could get to the point of having that combination model that we felt confident in.
I would like to thank NIAID, Judy and Ed and Freyja as well as others, the companies that provided vaccine, BEI Resources for providing us all of the reagents for our ELISAs and TNAs, a whole multitude of technicians and study directors and stats, QA at Battelle.
One person that isn't on here, too, is Jim Eastep that helped us design these initial models. He's not at our facility anymore but he should be recognized as well.
If you have any questions, I'll take them now.
PARTICIPANT: In looking at the Kaplan-Meier curves there seems to be a gap in the protection. I think the trick here is kind of a handoff between the protection provided by the antibiotic and that by the vaccine. From days 10 to 15 there is a little window there where you see deaths in the most efficacious groups.
It would suggest that window really hasn't been completely closed. I am wondering what the thinking was to stop at seven days the antibiotic given what you know about peak responses to zero to seven vaccination a nd so on.
DR. MOTT: The normal time of death in the control animals being 2.4 to four days basically. We went back and looked at the data of animals that were treated with antibiotics we did see a slight delay in the time to death in those groups.
Especially it seemed to increase with the increasing time of antibiotics so five versus seven versus nine. If you look back at that previous data there was an animal in that five group that was at 30 days that really expanded that group out to look like the seven days.
We picked that time of seven because of the significant death that we saw in that group as well as the time to death which we suspected based on previous data that showed that the accelerated vaccine regimen would provide us some immune response by 10 days. We were seeing times of death of 12 days, 13 days as an average and we suspected that would be a good way forward.
DR. SELF: In retrospect, would you think that a ten-day course of antibiotics would close that?
DR. HEWITT: I'll answer that. If we extend the antibiotics to 10 days, then the antibiotic only control group would have been a higher percentage of survival, more like 80, 90 percent. We would have had to have very large groups in order to get a statistical difference between the antibiotic only and the antibiotic plus vaccine.
I agree completely that there is a gap there and basically we chose the seven days because we were still able to demonstrate the statistical difference between those groups.
DR. SELF: I understand the arguments about study design, still the regimen to me is somewhere along the course length. Do you know what I mean?
DR. HEWITT: Right, but I think what you will run into then is that the antibiotics themselves will be so protected that you may not be able to demonstrate added value of vaccination.
DR. MOTT: In some of the data --
DR. HEWITT: I agree this is artificial.
DR. MOTT: Some of the data that you don't see in a previous study indicates that as well from that last figure where I showed we kind of took one way versus the other. An earlier study did have a longer administration of antibiotics that had a lot higher survival as well.
DR. SELF: So then a ten-day course prior to vaccination, wouldn't that be credible?
DR. HEWITT: Potentially, yes, in rabbits.
DR. CHAWLA: Just to add to what he is saying, usually most of the vaccines taken 15 to 20 days to really give the effect. I have been working with vaccines for the last 20 years and it seemed that the response of bacterial vaccines. It takes 15 to 20 days for the immune response to come so it will be a good idea to see the studies for at least 15 days.
My second question is related to antibody which is the antibody in human. Is it levofloxacin or ciprofloxacin?
DR. MOTT: They are both approved but for at least for our studies we have to go to the levo based on tolerance issues for the rabbits.
DR. CHAWLA: But, as you know, yesterday we discussed many times that we are focusing on human vaccine, not a experiment in animals. The focus should be use the antibody in human beings.
DR. MOTT: The antibiotic or the antibody?
DR. CHAWLA: Yes, antibiotic which can be used --
DR. MOTT: We do use the other antibiotic with nonhuman primates but for the rabbits based on the tolerance issues we couldn't use it for an extended period of time.
DR. CHAWLA: Is it going to be levofloxacin as the antibiotic of choice in humans in post-exposure cases in combination with --
DR. MOTT: I don't think I can make that call. That is not up to me based on these models.
DR. CHAWLA: Okay. My third question is that there is a lot of potential for exposure in terms of GA exposure so any plans to have a post-exposure model for GA in addition to aerosol challenge? Because it is easier, I would say, for anybody to expose human beings to GA exposure instead of aerosol.
DR. MOTT: I don't know of any right now.
DR. CHAWLA: Let's say the spores mix with water or food.
DR. MOTT: We have been focusing on the aerosol.
DR. CHAWLA: Because that could be kind of an easier route of exposure than creating aerosols, etc. A study should be designed for that kind of exposure also.
DR. FRIEDLANDER: Just a comment in reference to the design of the experiment and the question that was asked before. As I understand it, the purpose of this is to try to demonstrate that vaccination can enhance the efficacy of post-exposure so it is very dependent upon the animal model.
It is dependent upon the dose that you use and the retention in the particular animal, some of which is unknown and has to be determined empirically as was done here.
That is to say, how long do you need to treat such that when you stop you will get sufficient mortality. If you don't, you can't obviously demonstrate the added benefit of vaccination. That may different depending on the model.
DR. MEADE: In the back.
DR. BIGGER: John Bigger, Battelle. Jason, I just wanted to ask was there any indication that survivorship after cessation of antibiotics was immune response dependent? Was N bright enough to do a logistic regression on this like the day 10 ELISA data comparing survivorship to immune response?
DR. MOTT: Say that again?
DR. BIGGER: You know, Mark and I both presented survivorship based on immune response in our talks and in our case we had very defined control and a range of immune responses and our end was pretty high in our survivor versus death rate. You may not have that luxury here but maybe at day 10 if you were to look at the immune response in the animals at day 10 it might be a predictor of survivorship or not.
DR. MOTT: I don't know that we did that analysis.
DR. MEADE: We are going to talk about that a little bit in the next talk or, at least, explore that a little bit in the next talk as well as in the panel discussion. It's a question that is being thought about, yes.
In the middle, please.
DR. GLYNN: Kate Glynn, CDC. I found this model very intriguing and I think very useful as a first step. It clearly does show that the addition of vaccine on top of a microbial agent is beneficial.
I guess what I would like to her is how you think the next step -- how this model could actually stand up to the next step which is talking about how post-exposure prophylaxis regimen using vaccine and antimicrobial agents would over the estimated potential incubation period for spores actually provide the protection that we're talking about for PEP and whether this model could be used to extend.
I don't know if that may be getting at what some of the other people are asking as well. It shows very nicely this particular issue but whether this model could actually apply to the broader question.
DR. MOTT: I certainly think this model is shown to be very well characterized in reproducible. Do I think it could be used towards the next step, something that we can interpret the data here to be extended to humans? I do. I realize that not everything is optimal here to look like a human infection.,
DR. GLYNN: No, I'm sorry. Just to be clear, not to just look more like humans but the situation we are trying to address in humans if that model could be recreated -- I mean, that situation could be addressed using this model, the animal model.
DR. MOTT: As a deliberate release, for example.
DR. GLYNN: And the longer-term protection so this addresses short-term.
DR. MOTT: I don't know whether or not this would provide long-term protection. I would guess that it would. I know one issue with the rabbits that isn't going to be the same in humans is that the period to treat to keep this model going versus getting into an area where it's therapeutic is a very little window so I don't think we can extend it out much further.
Obviously with the primates they seem to have an extended period of illness by a couple days. I think once we get that going maybe we can extrapolate from that model a lot better towards what we would see in humans.
DR. MEADE: Pat.
DR. FERRIERI: Pat Ferrieri, University of Minnesota. This is very nice preliminary data. I gather we are not going to hear anything on the nonhuman primate along the same lines of combination of antibiotic with vaccine.
My interest among others is the pharmacokinetics of the antibiotics in the different animals. I'm sorry that your rabbits couldn't tolerate the ciprofloxacin because I think it is viewed as the preferable agent in the human situation over levofloxacin.
Levofloxacin human serum levels were much higher than the rabbit area under the curve so I guess it doesn't completely address the issue of whether the antibiotic would suppress the immune response.
Have you give any thought to that or maybe someone else in the room has data on the ciprofloxacin in the nonhuman primate that would address the issue of suppression of immune response. Nonhuman primate levels may be more comparable theoretically to ours with cipro or levo.
DR. MOTT: Well, having run a study with primates and the vaccines and antibiotic in combination, although the data isn't being presented here I can tell you that it is similar to the rabbits that we don't see any changes in the levels of the immune response. The antibiotic levels in the primates do tend to be higher with the cipro but we haven't seen any interactions that show it does inhibit the response.
DR. FERRIERI: One other quick question. I assume that the ciprofloxacin or levofloxacin does not interfere with the TNA, the ability to conduct the assays, the TNA?
DR. MOTT: I think I would give that off to someone in the middle area.
Kristen.
DR. CLEMENT: We looked at that during our validations studies, the effect of various types of rabbit matrices and one of them was rabbit serum from rabbits that had not just been -- it wasn't a spiked situation where we look rabbits spiked with levo.
We actually gave rabbits the highest dose that has been used on studies of levo and drew their serum to use in our TNA validation studies. The end result is there is no effect of the levo metabolites.
DR. FRIEDLANDER: Just to address this question that was raised about whether in the primate ciprofloxacin will interfere with the immune response. We published a paper last year on the nonhuman primate in which vaccination occurred post-exposure along with ciprofloxacin.
Those survivors developed an immune response and were resistant to rechallenge although we did not compare the levels to a group that just received vaccine alone. The antibiotic did not inhibit the ability of the vaccine to induce a protective immune response.
DR. MEADE: Okay. In this last talk before the discussion I wanted to -- I was asked to at least talk through some of the issues related to use of the PEP model in the area of correlates. Since the purpose of the meeting is sort of translational, how do we think about applying the PEP model to humans?
The question has already come up. There's a number of issues that have been raised and I think I just wanted to go through a few other points that we thought about or at least questions we've asked. I won't have any questions. Again, my job was really to raise questions and hopefully lead into some of the panel discussion issues that we brought up.
I certainly won't raise all of the questions and issues but, again, I did want to highlight a few of the complexities that we have thought about. A lot of these -- many of these come from hours of discussions with Drisilla Burns and many other people in these groups. This is reflecting the thoughts.
Again, just to focus back on the issue of the meeting in terms of the animal rule and, again, what we are focusing on primarily is the fourth item. The fourth issue is the information that will allow the decision to be made in humans and how do we begin to think about linking those models.
Again, I think this was raised or has already been brought up but, again, I just wanted to refocus and make sure that everyone is clear what the issue is that we are really talking about, the management of individuals after exposure or possible exposure to Bacillus anthracis spores.
Again, either some post event which I define as maybe or presumed event, possible exposure, or exposure. I don't really see that those are different. In most cases you're not going to know if there was an exposure or not so I perceive those as really the same thing but to distinguish it from treatment.
This is really not the intent to be dealing with a treatment after onset of symptoms. That's the models that were being designed to deal with those steps. I think there is no doubt that the primary approach will be antibiotics and the whole issue is to evaluate vaccines as a supplement to that. I don't think there has been any intent or belief that antibiotics would not be part of the approach.
Again, I think just to reiterate what I think was one of Judy's comments is the animal studies for post-exposure will build on the pre-exposure or general use studies we talked about yesterday, the active and passive immunization studies. They are really intended to build on all of the information that has been gathered and the really nice data that we heard yesterday.
Again, you saw a number of these curves yesterday where the general approach to coming up with some estimates of what concentrations of antibody would be protective.
Again, you saw a number of these curves where you have plot the antibody of the titer as a function for individuals whether or not they died or survived and then logistic regression or other models you can make a plot of the probability of survival as a function of TNA and then, again, with confidence intervals you can make the kind of modeling and come up with estimates that we heard about a number of times yesterday.
I will speak only for myself but when we were thinking about this at the beginning I sort of naively thought you can do this with PEP, too. Why not sort of think about the same thing. What's the problem? Go ahead and take the TNA and make these plots.
The more I thought about it the more I realized that is really hard and has more problems and issues. I'm going to spend the next few slides talking about what I think are the reasons that's problematic and needs someone far smarter than I maybe to help put this all together. Again, I think that will be the focus on some of the discussions.
Again, just to remind you, I'm going to build on the data that Jason has just presented but, again, just to simply it, we are talking about studies where there is a group that receives antibiotics alone, one that receives antibiotics and vaccine. You obviously have a control group that receives neither just to be sure your challenge is behaving properly.
After some minimal amount of time needed for the vaccine to elicit an immune response the treatment is stopped in the groups one and two and then you are going to monitor survival over time which, again, is the data that Jason presented. Then you are going to compare the group that received antibiotics and vaccine to that who received antibiotics alone.
I think I had group three in the handouts incorrectly thanks to Bob Kohberger. He noted my error. Again, that is sort of simple. What I've done here is maybe inappropriately from a purist point of view but, again, I was trying to keep things simple.
As I put together the three experiments that Jason just presented to keep it simple because the specifics matter less than the general format of the data. Again, just to highlight the basic data, those animals that receive no antibiotic at all and no vaccine die very quickly.
Again, those animals in blue -- I'm color blind so I try to do colors that I can see but I don't necessarily know what to call them. I'll keep pointing and hopefully the shapes will help. The blue line is the groups that receive the antibiotic with no vaccine.
Again, it just shows, as Jason said, that there are no deaths during the time when antibiotics are around and beginning a few days after the antibiotics are terminated the deaths occur over a fairly long window of time which again differentiates that from a lot of the studies you saw in the pre-exposure is that the deaths occur in a much shorter window of time.
Again, they are occurring from 11 to 28 or something like that. Again, the three groups. I pulled the three vaccines, the low dose, the mid dose, and the high dose to show the time at which deaths are occurring. You clearly saw added value from that perspective.
It clearly shows that the vaccine is adding value to the system. Again, my simple-minded thoughts at the beginning were the ideals situation as you would estimate the antibody titers and animals protected by the vaccine.
You compare those eventually to those levels achieved by humans receiving the vaccine administered on the schedule intended for the post-exposure prophylaxis. Then you would use approaches we talked about yesterday and sort of make some estimates about the efficacy of vaccine for post-exposure indication.
However, again, as I have said, as I thought about this estimation of a protected level from the pet model, at least in my mind, it's a lot more complex than with the pre-exposure models we heard yesterday. Again, I just want to talk through some of the ones that I've thought about. Again, I think we have already heard a few others that I think we can talk about further.
One is the fact that the pet model is a very dynamic situation. There are many things happening at the same time and I'll illustrate that. Again, you have an infection going on at the same time you have immunization.
Again, as has already been alluded to, and Jason mentioned this, too, I mean, the fact that you're immunizing in the presence of organisms I think there are questions that the antibody response could be influenced by infection.
Then there are certainly -- again, because of the nature and reality of this I think there is greater uncertainty in the statistical modeling, at least based on the sample size as we have so far and I'll illustrate at least what I mean by that as an nonstatitician.
Again, just summarizing the data just to give the shapes of the curves more than any details, this is the antibody response from the three PEP studies, pooled the TNA as a function of days.
There is a dose effective in going from low, middle, high but the more important things is the kinetics, that the immunizations are given zero or seven. Beginning a few days after that seven-day dose there is a very rapid rise of antibodies during the period of the study.
Again, just to highlight what at least my perceptions or uncertainties regarding the PEP model, the first issue is the fact that it's a very dynamic situation.
As you saw, some of the immunized animals are going to die following removal of the antibiotic and they die, at least the antibiotic only control group, died at a very wide interval of time from days 11 to 28. As I showed in the last curve, the antibody is very rapidly increasing during that same interval.
To relate, again, to the question John Bigger is asking is what TNA value should be used in some of the modeling to define some protective level or try to make some estimations of what TNA value is relevant.
Again, I think, at least my simple-minded view, for those animals that die you presumably could use the TNA value in the last sample prior to death. That is one option I think you could consider but I don't have -- I have far more trouble with what would one do with the animals that survive.
To illustrate to put the two graphs on the same slide that I just showed, for animals that survived in the challenge what point should you be using for doing this modeling? Again, I think one could make several proposals or, at least, we have kicked around several proposals.
One is why don't you choose day seven, the day the antibiotics are discontinued or maybe day 10 or the day of the first step, something like that, that phase. Clearly if you look at the antibody response curve, that is very early on. It's very low responses.
At least, from my perspective, you could be well beyond her estimating the value that is relevant during the window of time that matters. I guess you could also propose that you look at the peak or maybe the last day in the control group, something down here.
I think, again, that is very likely to be overestimating the protective values. Again, because it's basically going from all to nothing during that same window, you can sort of choose anything you want. I think that is where I'm based.
Choosing a logic for choosing a time or, at least, modeling a time, is something I think we hopefully will have some discussion about. I'm a little uncertain as to how one would make that and defend it in terms of a modeling point of view.
I think, again, the second issue that I think in the first animal model working group discussions on this, I think Jim Eastep raised this question that you are immunizing in the presence of an infection.
I mean, clearly there are antigens from the organism that could be there or the infection process could be influencing immune response in a positive or negative way. I think it's different. I think there is some evidence from studies.
The studies were never designed and empowered to look at the immune response in this way but I think there is some evidence that the antibody concentrations may be higher, at least in some of the rabbits, that receive the vaccine and challenge and others receive vaccine only.
I think to answer this question, if it's necessary to be answered, you look at the TNA titers in the two groups, one that had the antibiotic and vaccine and challenge versus those that got antibiotic and vaccine in the absence of challenge. There are some animals in that comparison.
Again, I think they certainly were not powered to look at differences but there is certainly some evidence that some individuals did show an increase that certainly could be attributed to the presence of the challenge.
The issue of three deals with statistical issues. I need to give a disclaimer that I'm not a statistician. I should have brought the picture but a statistical colleague of mine at my retirement event last winter gave me the book, "Statistics for Dummies."
I think there was some comment about my statistical experience in there. Again, hopefully I can at least illustrate the questions and hopefully will turn it over to statisticians who can turn this into more details.
The issue is that the survivors, at least using my terminology, represent true positives and false positives in that they are 40 to 50 percent of the unimmunized animals survive.
We had, again, looking at the animals immunized with antibiotics about half the animals are surviving, again, in the absence of a vaccine. For any given animal vaccinated that survives, can you differentiate whether or not it survived because of vaccine or because of -- or would have survived anyway.
I think what you're doing is population siding. You are clearly showing added value so that clearly shows added value vaccine. In terms of modeling I think you have the concern that half the animals will survive anyway.
Than, again, the other point, at least with respect and comparison to the pre-exposure studies we talked about yesterday, relatively few immunized animals in these PEP studies died. Again, we are talking this group here and I think I did the counting and there were 162 total animals in these groups of which 14 survived.
We are talking about relatively small number of animals -- excuse me, that died. I'm sorry. Fourteen animals that died in the immunized group. I think generally we are doing modeling using the responses in immunized animals so I think, at least at this point, we have relatively few animals to be looking at.
Again, from a modeling point of view I think, again, we saw these yesterday. In the pre-exposure models, at least in the rabbit, all the low responders all died and all of the high responders lived so it's relatively easy to -- you don't need very many animals at the asymptotes when it's an all or nothing event and you can really focus your studies on the range of interest where you are going from, the death to survival.
With the PEP studies we are really dealing with a lower asymptote in this 40 to 50 range. Again, my picture of it is graphed in here and if you are doing a similar curve comparing it to the rabbit GUP model, here are two curves which have the same E50 slope and asymptote. For the rabbit PEP model you are going to be fitting data from this kind of curve.
Again, you may not need that many animals here where there is the full protection but down here you are going to leave a lot more animals down here in this range here to really give any precision of your estimate of the values to come up with the shape of this curve.
Again, the fact that it is a much shallower curve, again, I think you generally need more animals and more data when you have a flatter curve and you have basically less of a range differentiating between 50 percent and 100 percent. You are going to need more animals and really have quite a bit more data in here to really come up with a model with any precision.
Again, I think, just to summarize at least my understanding of the challenges to statistical modeling is that because 40 to 50 percent of the antibiotic treated animals survived without the vaccine, there is greater uncertainty in us estimating the lower asymptote in E50 and to achieve comparable precision in estimation of a protective level PEP studies would need to include many more animals than GUP studies. I noted on Judy's slide they have already done 700 animals, I think, so far. We are talking a very large resource commitment.
Again, I think, in the studies done to date I think there's relatively few animals that died, immunized animals that died so future PEP studies, again, if one were trying to model, would need to include a wider range of doses including more sub-optimal doses and adequate modeling requires a sufficient number of surviving and nonsurviving animals from the immunized groups in order to do those models.
To summarize, the complexities in estimating a protective level from the PEP studies the fact that the PEP model is a dynamic situation which antibody concentration is rapidly increasing. Secondly, the antibody response could be influenced by the infection and then the fact basically of the nature of the model suggest you need more animals to do the modeling.
Again, I think just to go back one step, again, I think even to do this modeling -- one point I want to make is even to do this modeling one would have to make a decision about what, in fact, time point TNA you are going to model. I think that is in addition to the fact that you need more animals you would have to make a decision and come up with rationale for how one would do that.
At least from my view, defining a protective level for TNA based on the PEP studies it appears to be in my mind challenging resource intensive. I think given that I just really wanted to raise some questions and some thoughts that I think will lead into the panel discussion and, I think, what alternatives should we be thinking about if we are unable to obtain good estimates in protective levels.
I think one of the things that we need to keep in mind is that as the studies proceed, data is going to become available for different vaccines, different species, different models, the general use of pre-exposure, the passive protection, post-exposure.
I think one question to ask in this context is really how -- I guess my approach to PEP, or at least one approach to PEP, would be how should the data from the various models be integrated to make judgments regarding the value of immunization in the PEP application.
I think the question merits discussion. Again, there are some very bright people in the room that may have some thought about this and have some other interesting idea but can a protective level be defined for the PEP model.
I think importantly it should at least be asked is it necessary to divide a protective level for the PEP model. Is that necessary. Again, I think where we are now, what other data should be gathered in PEP studies. And, at least from one perspective, is it reasonable and appropriate to assume that a TNA determined to be protective through the pre-exposure GUP study would be protected in the PEP scenarios.
Or is there any reason to believe that if it's protective in one it wouldn't be protective in the other. Hopefully that will be discussed. I think, just to conclude, I really wanted to come back to a reminder of how the studies were designed is the PEP were really designed to achieve the primary goal of demonstrating added value of vaccine.
Can you, in fact, demonstrate and certainly in the rabbit model as designed has clearly shown that with a lot of work involved as Jason and Judy have shown it was very difficult to get that. I have done a lot of animal modeling work in my life and I think this was more complex and more difficult than anything just because there are so many variables and such a fine tuning required.
It was a very difficult but, again, remarkably reproducible at this point and clearly shows the added value of the vaccine. I think at least one approach is to emphasize the data from the GUP studies when defining TNAs that can be used to predict protection as opposed to trying to do it in the model that may be very complex.
Anyway, I would be happy to respond to any questions at this point.
DR. CHAWLA: Anil Chawla. The causing mortality in rabbits in terms of LD50 and in terms of how quickly it can kill rabbits, will it have the same kind of mortality in LD50 monkeys also or nonhuman primates?
DR. MEADE: I'm sorry?
DR. CHAWLA: The strain of challenge, the challenge strain that causes mortality in rabbits, let's say one species of animal, will it have the same kind of mortality or LD50 nonhuman primates?
DR. MEADE: I think I can safely say I don't know. Again, I am here from a data analysis perspective. I think there are other people in the room that have a lot more information about the strains and the models but, again, I think this was really focusing on what has been developed so far which is the rabbit model.
DR. CHAWLA: Can somebody from the audience answer this, the LD50 in rabbits and monkeys the same, the mortality the same from a particular strain?
DR. MEADE: I think the question is is the LD50 for this particular strain at least similar in rabbits and nonhuman primates.
DR. PITT: Similar.
DR. MEADE: Similar is the answer from Louise Pitt.
DR. CHAWLA: There is no need to adapt this strain to a particular species, the strain which is killing rabbits very quickly like within 48 hours in 200 LD50 need to be adapted in monkeys before it is used as a challenge strain?
DR. PITT: No.
DR. MEADE: Larry.
DR. WINBERRY: Larry Winberry with Biologics Consulting Group. Enjoyed the talk, Bruce. Good to see you.
DR. MEADE: But --
DR. WINBERRY: No, no. Actually, we're on the same page in terms of being statistically impaired. In looking at the modeling it would seem that from the vaccine doses that were used in the study it would probably add value if we looked at a more fractional dose to get a lower protection. It seems weighted towards the higher end of your survival curve.
As you indicated, there is a weakness in the lower asymptote whereas if we were to perhaps exchange one of the higher doses which seem to have less discrimination capability for a third dose that perhaps was lower, we could fill in some of that data as well.
Then the other point would be it's problematic, as you indicated to identify where you take an antibody or TNA level as a predictor. I'm wondering when I looked at the antibody response curves, although there is a difference in peak height, you also see a difference in obviously the rate of change for the different doses.
I'm wondering if instead of an absolute cutoff one looked at, say a velocity or a rate of change in antibody titer as a predictor of survival since that seems to be dose dependent as well.
DR. MEADE: I think two comments. One is in terms of the kinetics. In order to really fine tune the kinetics you would probably have to take more samples. You also have relatively few data points.
When you get a slope the question is is it the time which immuno response begins and the slope and the velocity is the same. There are multiple issues that require a lot more fine tuning. At least my perception, and I won't speak for anyone, is that the status of the PEP model in the rabbits is the groundwork has been laid for some of the more definitive experiments.
A lot of work went into this to show that it is reproducible, predictable, definable. Then I think the definitive studies are obviously going to be discussed at length between what are the final and pivotal studies and I know I won't be part of them.
DR. WINBERRY: I was very impressed with the work that was done by Jason and his colleagues. Excellent studies.
DR. MACALUSO: Tony Macaluso from BARDA. I'm curious if these and other studies result in a PEP indication, I'm wondering how narrow or broad that indication will be. Will it apply only to cipro or to drugs in the same antibiotic class or to any antibiotic? I think this is a real concern as we think about possibilities of events with multiple drug resistant Bacillus anthracis.
DR. MEADE: I will choose not to respond because I know I don't have that information but I think that is the kind of question that should hopefully continue to be raised during the discussion. I think that are some of the issues that need to be further discussed.
DR. SUTER: Very simple question for you. When you challenge these rabbits and you put them on antibiotics where do you actually see the Bacillus? Is it still in the lung or is it transported somewhere else?
DR. MEADE: Is there someone from Battelle that would choose to respond to that? I think the question is where are the organisms following the infection. Can you speak into the microphone?
DR. MOTT: The bacteremias taken during these studies are from the blood itself. We also will sometimes do tissues if needed --
AUDIENCE MEMBER: Can't hear you.
DR. MOTT: The bacteremias that we take during these studies are from blood samples taken from ear veins, for example. Also tissues at the time of necropsy if needed to show bacteremia but we don't specifically look in the lungs post-challenge through lavages or any type of direct sampling that way.
DR. SUTER: I think it's interesting to compare rabbits with monkeys, for instance, because the anatomy of the lung is totally different because you have balls which you do not have in monkeys. I think it would be important to see how the processing of these Bacilli are occurring between the different animals because you might see differences in the immune responses between rabbits and monkeys.
DR. FUNNELL: Can I just --
DR. MOTT: Sure, it would be interesting. We just haven't had a chance or opportunity to do that based on the design of these studies and the main goals at this time.
DR. FUNNELL: Can I just follow on from that discussion? You mentioned in your talk that a fluctuation in temperature and an increase in CRP was something that you found in the rabbit studies. Can anybody comment about those observations in NHPs?
DR. MOTT: We do look at those parameters in HP protocols as well and we do see changes in blood parameters. The temperatures may not be as good in the primates but we do see changes in the CRP as well.
Typically it's going to be baseline values for hematology as well as CRP before challenge but post-challenge or at times when these animals show illness you do see changes in those parameters that we've used to help develop the models as well.
DR. KELLER: James Keller, FDA. I don't know anything about anthrax so perhaps I shouldn't be asking the question but it seemed like in a post-exposure situation a lot of the complexities for active immunization would go away if you used a immunoglobulin combination therapy with antibiotics. Was that addressed earlier?
I didn't catch the reason why we are going with active immunization and a post-exposure instead of trying more directly. I think a lot of complexities would go away with animal variation, variability to immune response if you gave a single injection of an immunoglobulin to the PA.
DR. MEADE: Again, I know I'm not the right person to respond other than I know that I think many options are being considered. I think those studies evaluating all of the options for treatment and vaccine options and I think the status is to collect appropriate data that allows judgments to make.
I think others will make some decisions involving practicality, cost, and a lot of complex issues which I think will ultimately have to be taken into account. It is a broader level question that I think we can focus today.
DR. NASS: In response to that question -- I'm Meryl Nass -- there have been several companies that have developed monoclonal antibodies for use. One down the street in the small rodent model and showed that the use of the monoclonal was effective and also served as a vaccine so that when animals were rechallenged six or 12 months later they were protected.
There also has been polyclonal serum collected from a number of military vaccines that has been stockpiled by CDC so that there is some ability to treat using that. In China polyclonal animal hyperimmune serum are used to treat humans who develop anthrax.
DR. MEADE: Thank you. Okay. I think I will turn this over to Freyja.
DR. LYNN: We're making up for all the extra time we had yesterday by running late this morning. I think in the interest of efficiency and good use of time why don't we go ahead and take a 20-minute break now and that way we can come back after the break and have the discussions and the wrap-up. It is now close enough to 10:00 o'clock. Please reconvene at 10:20. Thank you.
(Whereupon, at 10:00 a.m. off the record until 10:17 a.m.)
DR. NUZUM: Okay, so we are going to start our last session now. The agenda has changed a little bit. We are going to combine the two discussion sessions, one for PEP to conclude the previous session and then we'll have a final wrap-up, general discussion.
These are the panel discussion points for PEP. They are very similar to what Bruce had concluded with. I thought I would go over a few notes just from the questions that I took.
Maybe just to start the discussion, there was one on the challenge route, why we do aerosol. Keep in mind these studies are funded with biodefense dollars which means we have to develop products for biodefense indication.
As Drisilla pointed out, the animal rule requires that we do studies that support the label indication, and the biodefense threat is considered to be aerosol so we do aerosol challenge studies.
I think there was a question on why we are doing a vaccine for PEP or what the indication and so forth is. That largely has to go to compliance. I think the license regimen for antibiotics post-exposure is 60 days. I think in the last incident downtown they actually prescribed 120 days. I don't think any of us thinks that people will take antibiotics that long.
The other component of that is antibiotic -- just from the standpoint of how much is in the inventory. If antibiotic duration could be shortened. The license regimen could be shortened and the antibiotics will go further. Those are a couple of reasons why we are pursuing this.
There was a question on long-term protection. This is probably addressed in Bruce's slides, but I think the next step for this is fairly straightforward. This is where we combine the GUP data where we have a sense of what protective levels are in a post-event scenario.
If you take an antibiotic and the vaccine and then we do studies, we'll know what the antibiotics in humans are. We can relate that back to the efficacy data we get from the GUP studies. I think it's going to show people are protected in the long-term. There is really nothing left to do. I think that should fall out pretty well.
Kind of along those lines Bruce made a point post-event versus post-exposure, and he said they are essentially the same. I certainly agree with him that they are the same in that we almost never will know if it is truly post event or post-exposure.
However, I think we need to keep in mind they are very different scenarios. From a regulatory label point of view, yes, it's post-exposure because that is something you can clinically address. You can clinically test and we can do studies to design that.
In actual fact we won't know if people have been exposed or not. In that case the duration of immunity is important. I think we just have to keep in mind -- I think my main point there is that they are different and because when we are designing studies and the questions we ask in the studies are vastly different.
As with yesterday there is clearly a desire expressed to have NHP data and there are questions about the antibiotic used. We definitely agree we need an NHP PEP model. We will do Cipro in NHPs and that should help address the antibiotic question.
Keeping in mind, and I know it has been said several times already this morning, the purpose of these studies are to show value added of the vaccine. It's not to emulate the antibiotic used or the antibiotic regimen that's used in humans and rabbits. That is not the point. The point was to get a model where they are partially protected by antibiotics so that we can show the added value of the vaccine.
That said, going back to my point yesterday that these animal studies to support the animal rule are really about people. Even in these studies we try to use -- and if they are refined and we learn more about the GUP, the dosing and so forth, we will come up with what I refer to as a humanized vaccine dose such that the immune response elicited in rabbits is similar to the immune response in humans.
Similarly the antibiotic regimen we use in these models will be based on blood levels that don't exceed blood levels expected in people. That is kind of where we have -- we have to design the model to answer the question we want, but in so doing we keep in mind the human clinical situation.
There was a question on the antibiotic that will be given with the vaccine. Remember that whenever you start talking about antibiotics post-exposure you are talking about CDER. When we are talking about vaccine use it's CBER.
So our emphasis is just on the vaccine at this point and presumably once it has shown value and an indication is approved for PEP, that will simply be added to the current license. This is my opinion. This is a very regulatory oriented issue so I can be wrong and people can correct me, but this is my look on it.
What we are doing now is to get the PEP indication for the vaccine that will be simply laid on top of the license regimen for antibiotics. If there are other studies down the road to change the antibiotic regimen, the post-exposure antibiotic regimen, that will then involve CDER.
It will probably another round of studies and discussions and so forth. I see it as a staged effect where we bring the vaccine in post-exposure. If there is a need to shorten the life of the regimen, that will be a whole other effort.
The vaccine, I would presume, would be able to be added to any of the antibiotics in the stockpile. I mean, we know very well their effect on the organism. All we need to show is that we can add the vaccine to any antibiotic or one or two antibiotics. I think at some point it would be safe to say it will work with any antibiotics.
Okay. I think that is all of my notes on the PEP. Let's start with this first bullet point and that is to comment on the soundness of the design of animal studies to demonstrate the value added of vaccine for PEP indication. I'll do as Drisilla did yesterday and just open it to the panel if anybody would like to comment.
DR. LYONS: Yes. I just wanted to congratulate Judy and her team and Battelle and their team for doing a fabulous job. I mean, I don't know how many people can appreciate it, but these are extraordinarily difficult studies, and to do it as well as they are done you get lots of kudos there.
I think there is some confusion a little bit out there. Just remember this is a very artificial situation that they created to ask the question do you get an enhancement or not. Normally, as Steve was noting, people wouldn't be treated for seven days and then be vaccinated.
Those kind of things wouldn't happen. In order to see it statistically, you have to create the artificial situation to see it, so let's be clear on that. That is what the design is supposed to do, bring out that phenotype of enhanced protection by the vaccine in this artificial setting, and it did very well, I thought. I mean, I thought it did that.
I think where a lot of the -- and then it comes a little bit to making that leap of faith. I think that is what everybody is struggling with to a certain degree. Some of the questions I heard had to do with things that we are very unlikely to ever be able to know like this question of where is a spore sitting.
The real question is what is the reason for the breakthroughs in nature. Why do people sit there with the spores in the lung and then it breaks through. Well, it's probably due to both host events, the genetics of the host, and certainly some of the pathogens in trying to make some prediction about that rare event after long-term treatment with antibiotics is going to be probably not in reality statistically ever approachable.
I think we need to move on beyond those kind of things that are very important scientific questions that people will be working on a long time. But related to product development it is almost impossible. We just have to accept that fact that it occurs and move on. I think this design is an outstanding design and really brings it out.
DR. NUZUM: Thank you.
DR. FERRIERI: I'll comment next. I think the design is quite good in terms of our understanding of the rabbit model with set parameters, and it's a good starting point, but I think we need a lot more data and it may or may not be feasible to answer the questions you want answered. I think it's really tough.
The chaos of the immune response under conditions where you have antibiotics on board and you have the vaccine and some of the animals are going to make more antibody than others. I think it will be a very random event whether it's the rabbit or the nonhuman primate. I think we have to accept that we are not going to be able to control things as we would like to do.
Ultimately I think the model can be transferred to nonhuman primate. It would be great if we had had more data to see data, but I understand the limitations.
Again, it's going to come down to the numbers and how can you maximize this part of your question and the design of the study to the nonhuman primate without wasting animals unnecessarily and can you strategize by maybe doing a few studies to zero in on the correct dose, for example, and capitalize on the data that has been presented with the rabbit to try to streamline the studies in the nonhuman primate expanding the numbers of those animals in a way that can give us more precise information on the immune response as best as we can determine it with more than one event going on, the immune response to the challenge, to the bacterial challenge that was given, the spores that were given as well as the vaccine immune response. I think it is very complex, as you have all stated, but I think we need to streamline it.
I do feel these are valuable studies so I disagree with the member of the audience who doesn't think there is an indication. I think we should be prepared. I strongly support the efforts, and I want this to be part of the public record here. I strongly support the effort to come up with a reasonable approach that will give us added protection to antibiotics. I think this should be done.
DR. RUBIN: I would like to second the plaudits that were given for the execution of the study and the comments that Pat just made on the desirability of having this ability to combine antibiotics and vaccine in a situation where a real serious need arises.
Starting at that point I would like to highlight a complication of the animal model that was not explicitly considered. I would like to just develop that a bit. As you saw, if you take these animals with antibiotics, as you go on the mortality of these animals without any vaccine decreases steadily. By nine days the mortality is 71 percent and certainly we know that if we had treated these animals further, at some point of treatment their mortality would have been nil.
You ask yourself what actually did you do when you did that. The answer to that, at least in my mind, is that basically what you did is you slowly decreased the challenge dose. All of these spores that were inhaled at first, 200 LD50s, were sitting there and occasionally one would rear its ugly head and germinate and would promptly be killed by the antibiotics.
Slowly there was a decay of the existing spores so that you could actually say, with not much confidence, but you could say that actually at 10 days the animals really had a chance of only 1 LD50 because only 50 percent of them died.
I think that if you keep in mind that the challenge is markedly altered during this process, it becomes even more difficult to interpret the immunological data in a sense of what do you need to protect. And in my own mind, I think these two processes should be separated. I would recommend that they be separated, that we get good experiments to tell us how long do you have to treat an animal with antibiotics to have such a probability, one or another probability of them surviving.
From the other studies that we have discussed yesterday we get the data of what is necessary in order to protect an animal irrespective of preceding antibiotics and how fast that response can be achieved.
DR. SELF: So it was very helpful to hear that this is really a test of concept sort of model, but it does raise the issue that we are not now just trying to bridge across species, but we are now also adding another bridge that we are building, a completely different regimen, a completely different nature of exposure in some ways.
And all of this in the context of what seems to be a very complex temporal sequence of events. We didn't hear much today about the complexity of -- the underlying biological complexity of that sequence. We saw some kinetics.
The comment about decreasing dose was really interesting, and it raises some questions in my mind about the potential for thinking about modulating the challenge dose maybe in some experimental designs. I'm not quite sure what that design would look like but it raises that interesting possibility.
But I think, as I understand it right now, the potential for formally identifying some sort of intermediate outcome or surrogate and using that to motivate what would be a regimen, a combination regimen that would be used in people, that seems very remote to me right now.
It strikes me that an important test of concept is made in rabbits that can and should be replicated in nonhuman primates, but I'm having a hard time seeing how there is going to be much more precision in characterizing the added benefit of a vaccine that would be relevant to the human setting.
It may be at the end of the day that the efficacy of antibiotics as it is delivered in humans would be seen as the basis that there is very plausible, based on these tests of concepts, added benefit from the vaccine obviously weighed by whatever sort of adverse experiences vaccination might give.
But that would at the end probably be the ultimate basis for any sort of recommendation for a combined vaccine. I'm just trying to be very practical about the limitations of this model in this complex setting.
DR. RUBIN: I appreciate all the comments that have been made so far and basically agree with all of them. It's a very complex situation, and specific suggestions really are impossible in this kind of context right now, but maybe some overview comments may be helpful, overview comments about statistical approaches.
Much can often be clarified in complicated situations by embedding what you observe in data in a missing data context. That is, when you have an observed data set think about what you wish you had that would answer the question. It is a conceptual thought experiment.
Also it turns out that there has been tremendous progress made in the last few decades in dealing with those kinds of models and dealing with models with a tremendous amount of missing data. These models have made their way into astrophysics, into bioinformatics, into medicine, and so reviewing missing data very broadly.
In fact, there are some single papers in statistics that when you Google them they've had millions of hits that have been written in the last few years. That just indicates the, sort of the generality and popularity of these approaches. Maybe one example to just indicate.
If, for example, we wanted to think of these randomized experiments where you're randomizing dose of yesterday, but you sort of wish you had been able to randomize immunogenicity which you just can measure.
You can figure that in a missing data context in terms of noncompliance which is another term that came up today. There has been tremendous progress made on noncompliance in the last decade in statistical analysis with noncompliance.
What you can think of as is you wish you could randomize immunogenicity, but you can't or you didn't, but when you gave a high dose, you were trying to encourage high immunogencity. And so rabbits that were given a high dose and had relatively low immunogenicity were sort of noncompliant in that sense.
Rabbits that were given a low dose and had high immunogencity were sort of noncompliant in that sense. They didn't follow their encouragement. Those kinds of models have been used now for a decade and with great success in a variety of situations.
Although I agree with Steve that it seems to be a hopelessly complex situation because there are so many different features in this PEP environment, although I'm not sanguine about it, I'm more enthusiastic I think, perhaps, about the possibility of making progress if the problem is laid out correctly first and then see what you are missing.
Because very often when you do lay it out in this missing data context you see there are some key pieces that are missing, but they are not so hard to get by doing the right experiment. Until you lay it out seeing what all the pieces you need are, it's impossible.
You are sort of thrashing around so it's like you wish you had this and wish you had this. You don't have the full picture yet so ideally we'd have a meeting with people who understand immunology, which I don't, and some statisticians. Maybe we would lay out what we are missing and what we would like to do and maybe even discover that some of the missing pieces can be discovered in relatively simple experiments.
DR. FERRIERI: I would like to follow up on Don's remarks. There is a wealth of fantastic expertise in this audience. What I'm missing, Ed, right now is the plan moving forward who is going to be doing the nonhuman primate studies? Will it be one of the sponsors? Is some of it being done by NIH or FDA at Fort Detrick? Where is this work going to be done?
Is everyone talking to each other to plan the most precise experimental studies so the next time we may convene here the presentation of the data will be done in a way that is laid out comprehensively and in a manner that will permit at least me to digest it more easily?
DR. NUZUM: I would comment on that. Certainly NIH is planning to move forward with the NHP studies as we've done with the rabbit studies. However, it is getting more complicated as there are more funded companies developing multiple products, monoclonals, polyclonals, vaccines. It is an area that we need to look at in the government.
There's a lot of discussion on that very point. I kind of alluded to it in my talk that is where we need planning to maximize resource utilization. That is a very good point.
DR. SELF: This is a bit of a tangent, but it might be a good time to interject. Yesterday there was a lot of data presented but not all sort of analyzed in the same day and aligned.
I guess one of the things that I would encourage you to do would be to set up some process for maybe an independent statistician, maybe Bob or maybe somebody else, to get those data sets, analyze them in a consistent way and bring those results together as rapidly as possible. There's a lot of, I think, knowledge that is there in that data that hasn't been extracted yet, and it would be very simple to do that.
DR. NUZUM: Thank you. Good suggestion. Maybe we probably should move to the next bullet. Very good comments. The next bullet is take into account the complexity of the PEP system.
Discuss whether protective antibody level can be estimated from animal PEP studies that can be extrapolated to humans. If so, how might this be done? If not, discuss the strategies for linking animal protection data to efficacy in humans for the PEP indication.
Any comments? It sounds like there is general agreement that everyone acknowledges this is a very complex model. I guess my feeling, and certainly Bruce was very clear, but there is a lot of other data besides death we can look at, bacteremia, temperature, clinical science, other endpoints. I think there are ways to strengthen the model in the absence of a clear-cut correlate is my opinion. Would anyone comment?
DR. FERRIERI: I would just kick it off and see how people react. It's not going to be possible in my opinion to have a very precise understanding of the protective level in any of the animal models and be able to extrapolate it to humans where the complexity of the immune response is going to be even more unregulated, if you will, or unpredictable, whatever language you wish to use, randomized, etc.
I think we need to be very clear that there are limitations here. We accept those limitations and that we don't drive this into a hole continuing to expect more than is possible from the model in terms of having "the antibody level," the TNA precise level.
A term I use in clinical medicine at times when I'm seeing certain complex patients. We have to guesstimate. Perhaps the best we can do is that rather than something very precise. There will be a range, and maybe that will be acceptable.
DR. LYONS: Thank you. Ed, one of the things that I was thinking about, at least for those of us who trained in the older days, I mean, when one was exposed to hepatitis and hadn't been vaccinated you were basically treated and vaccinated at the same time. I mean, at least, if I remember correctly that was the plan.
It would be interesting to go back and see where that strategy actually came from and what they used to try to -- what was their correlate for protection in that situation. I don't even know if there was one. It might have just been it doesn't hurt to just do it all and it was done in sort an off-label basis.
I don't know. The precedent for doing this post-exposure is there in other diseases, and hepatitis B is probably the one I'm familiar with. And so it might be worth going back to look and see how they thought about that process.
DR. SELF: I agree with the comments about not getting too precise, but it does occur to me that if in the context of this model you build the same sort of logistic curves relating survival to TNA levels or whatever your favorite antibody measurement is to see if those even approximate those curves in the gut model, that would add a little more comfort at the end of the day even if not much more precision to calibrate what the effect would be ultimately in humans.
DR. FERRIERI: I agree, Steve. I'm not trying to convey that I'm not going to apply rigor and try to squeeze the most out of all of this. I have a high standard and high expectation of all of you who will be doing the work.
To follow up on this point about making corollaries with the hepatitis model, we have to remember that this is an overwhelming infection. Hepatitis doesn't get into the lung causing acute inflammatory response. The whole pathobiology of the infection is quite different with anthrax in the rapidity and the damage that is done.
One has to -- you have to intervene rapidly. Hence, the necessity of having the combined approach, in my opinion. The schedule has not been commented upon, and zero and seven days were used in the rabbits. I don't know what the plan is in the nonhuman primate, but I would assume it is somewhat similar.
I think zero and seven sounds good to me. I wouldn't be playing around with longer stretches necessarily of the possibility in humans of revaccinating again after seven days to perhaps shut down the need for continuing antibiotics for three months, four months, six months, whatever.
Consideration should be given in the NHP model to perhaps continue in the survivors to do that to boost the immune response. I would be interested in reactions from those who will be doing the studies.
DR. NUZUM: You mean boost with another vaccination longer out?
DR. FERRIERI: Perhaps three weeks later to do it again. Zero, seven, and then perhaps three weeks later or whatever. Has that been considered?
DR. NUZUM: Well, not at this point, but it is a good point. Again, I guess I would say we are kind of still in the basic model development. We have what we think is the basic model, and certainly there is room for tweaking and refinement.
But at the end of the day, and I think it was mentioned yesterday, the human regimen will depend on human data. We can look at as much of that as we think is practical in animals, but if the human response isn't the same as animals, then that is going to drive the final choices.
DR. FERRIERI: Of course.
DR. HEWLETT: We are separating the post-exposure from the general use, I realize, for specific reasons, but I want to go back to what Emil said. I think that we're talking here -- even though it's animals and not humans, we are talking about -- there is no doubt that antibiotics work as long as you keep taking them.
There is no doubt that the immunization works depending on repetitive onset. It seems to me that we are talking about protocol optimization.
Looking at those and getting the best -- especially since there wasn't any indication of interference, at least in the studies that were demonstrated, optimizing those protocols ahead of time and then combining them because you have so many variables right now if you do it that way and then put them together in a way to show that in fact -- because the idea obviously is to minimize the duration of antibiotic use and to optimize the rate of onset of the immune response. Those can be worked out individually.
DR. NUZUM: But it seems like that's what we did. We did do separate antibiotic-only studies and then used the information from the GUP studies to select vaccine dose. So, I think I am missing something.
DR. HEWITT: Judy Hewitt. I can add to that. I think maybe another sort of perspective comment on that. The pressure to develop a post-exposure model several years ago was so great that we felt that we couldn't spend the time to carefully to optimize both of those situations, so we sort of moved forward in a way that sort of got us to good enough as quickly as possible.
Having said that, you know, we dropped the start-time study in the very beginning. We just chose a start time. We are in the process now of going back and looking at that, did we pick the proper start time.
That data is not ready to present yet, but we are going back now and looking at some of that optimization, those kinds of studies. I think it's not that we are ignoring that. It's just that we sort of had to -- there was too much pressure to get there too quickly.
DR. HEWLETT: I understand those constraints. I think this takes me back to a question I wanted to ask a little bit earlier. We are talking about the ideal and the practical, and you are dealing with both of those at the same time.
I've gotten progressively confused as we went through all of the data here being asked questions like what additional data are needed when I don't really have an understanding of what the endpoint is. I realize licensing of vaccine and knowing the ideals, but we are not going to come up with a V&A cutoff number.
We are not going to come up with any of those specifics from this information, and I'm trying to get a sense of what it is that we are trying to get to by asking the questions of what additional things do we need. Do you understand what I mean?
DR. HEWITT: Absolutely. One example when the question was first brought up, what do we wish we had, we have, for example, in the PEP studies on the survivors, the rabbits that survived post-challenge, we have weekly ELISA and TNA values, so we can't really measure the kinetics of those responses on a scale any shorter than week by week.
I wish I had interim data points between those weekly samples. Sitting here and having this discussion brings that very clear to my mind that maybe we need to do a study where we look at more of those interim data points.
We don't need data points out at 28 days and beyond. Let's get some more data points early on in the response. I mean, I think there really are some good pieces of data that we can sort of focus and try and get next.
DR. HEWLETT: But how much? Talking to the regulatory people, how much do they need to have because we're not going to set levels here and now. We just want to be adequately prepared when a licensed application comes so that you can judge properly where it is. Again, this is the combination of the practical and the ideal. That is the intersection point.
DR. BURNS: Exactly. Drisilla Burns. What we want to hear from you, the panel, from the audience, is to get scientific input on how meaningful you think these models are in regards to being able to support efficacy of a vaccine. You have seen the model presented as it is with the types of data that are being generated.
So the endpoint is can the model, in your viewpoint, support efficacy. Any given vaccine could give a different result in the model, but the model itself, how good is it, how strong is it. Do you need something else in your mind that would be -- and we could do the need to know, nice to know sort of -- two categories. Does that answer your question?
DR. GOTSCHLICH: Yes. May I ask you another question? Why is this being separated into a different box? Let's just assume for the sake of argument that we have an effective vaccine. Why do you need a different box for an effective vaccine after a person has been exposed to anthrax and in order to be certain that that person did not get the disease was treated with antibiotics? Why do you need a separate box?
DR. BURNS: Well, I think in general it is a different indication that would go on the label.
DR. GOTSCHLICH: Why is it a different indication? The indication is to prevent anthrax in a person. You have two treatments.
DR. BURNS: And one indication is you are exposed after you get the vaccine and one is before. Now, I would love to hear any comments you have.
DR. GOTSCHLICH: I'm still not clear why it is a different indication. You have a human being that requires the vaccine.
DR. BURNS: Yes.
DR. GOTSCHLICH: In one case prospectively, in this particular case, there is an additional reason why that human being requires an effective vaccine. Why is it a different indication?
DR. BURNS: I think that we want to hear your thoughts on why you think it's not a different indication. We would like to hear your thoughts. What is the science that would say that if you have protection before exposure you are definitely going to have protection after exposure? I would just like to hear your thoughts on that.
DR. GOTSCHLICH: Well, I will repeat them very briefly. Basically it seems to me physicians have two things in their hand. They can treat the disease with antibiotics or the prospective disease with antibiotics and that is fundamentally slowly eliminating the challenge dose, or if there is no preexisting challenge dose in them, give them vaccine, and if they are challenged they will have enough antibodies aboard to be able to handle their challenge. I see these as completely separate processes.
DR. CLIFFORD: Julie Clifford, FDA, CBER, Office of Vaccines. I'm not sure I'm understanding exactly the question, but I'm going to offer something that might help.
In terms of the vaccine or any vaccine for anthrax, the indication and usage section of the labeling, once it's approved, will say that it offers or provides protection after a certain number of doses based on the clinical data or the animal model, you know, the combined licensure package.
But in a post-exposure setting to someone who is naive to vaccine, the chances of them achieving what we would hope to be identified as a protective level of antibody or an ability of their immune system to respond enough to protect them may not occur until after for this disease, in particular if they have already succumbed. That is why it would be seen as a separate indication. The preexposure indication would require a certain number of doses based upon clinical and nonclinical data to support licensure whereas a post-exposure we are still trying to get a handle on how you identify the number of doses and schedule in a post-exposure setting and how somebody could be protected.
Right now we have a separate component of antibiotics that affords kind of an immediate protection, and we are struggling with how to define the use of a vaccine that would add an added component in conjunction with the antibiotics. Does that help?
DR. GOTSCHLICH: Thank you.
DR. CHAWLA: I will just give an example. Maybe it can help. We have a rabies example where the preexposure indication is something different. You follow a different immunization schedule. And in post-exposure where it is human with immunoglobulins or without immunoglobulins in post-exposure you follow different kind of schedule.
So it is -- if a vaccine has to be shown efficacious in both the conditions preexposure or post-exposure, a different set of animal data or efficacy data is required.
DR. NUZUM: I had made -- and I skipped over because there was a question earlier about incorporating either monoclonals or polyclonals in these models. And that is being done. Remember again, and it was shown, there is preexposure, post-exposure treatment. We are not talking about treatment models. Those are also being worked, but it is a whole different data set you start with. It's a different model set up and endpoint. Yes, that's being done but, you know, we're not talking about treatment.
You have been standing there a while. Go ahead.
DR. FRIEDLANDER: I would just reiterate what Rick said that this is an excellent approach in beginning to address this important issue. I would also reiterate, I think, what Emil said. In its most simplistic sense I think you could make the argument that you have made, what is the difference between pre and post-exposure.
Why not just consider that you are using the antibiotics only to get the individual to the time when you have already established in a preexposure model that the vaccine is effective.
Assuming or demonstrating that the antibiotics do not impede the development of the immune response, you basically use the same regimen you have used or you could modify the regimen but use the same regimen you used as an indication.
What are you doing when you treat -- when you vaccinate post-exposure? You are not using the vaccine for the immediate exposure. You are using the vaccine for the exposure that occurs once you stop the antibiotics.
If you know in your mind what that time is when you develop a protective immune response, that is the time that you need post-exposure to be on antibiotics. I think you can argue that case.
DR. NUZUM: So, very simply, I think I would summarize this by saying if the vaccine is shown to be safe and effective on its own, antibiotic is shown to be safe and effective on its own, and then you can show that combining the two don't provide any adverse effect, what is the problem with using both of them? But then I think we are talking about clinical trials, not animal studies, to really show in people that there isn't an adverse effect.
DR. FRIEDLANDER: Absolutely. You would have to show, I think, that it's safe and that it doesn't impede the immune response. Those are fairly easy studies to do.
I would like to just make a couple points that came up before in terms of this business about the different models and the duration of antibiotics because, first of all, we know they vary in different animals, and there are functions, as we said before, of the dose, of the challenge dose, and of what the determinants are for germination or retention of spores in different species.
We don't know that. Anecdotally you can think they differ in the rabbit and the nonhuman primate, but we don't know that. But this idea of finding a measure of antibody that we believe confers protection is, I think, very important because if you knew that, you could tell someone when they could go off antibiotics if they had been exposed and, say, developed some infection.
Or if they were vaccinated, or if they were treated for established disease and developed an immune response you could say, "You don't need to stay on antibiotics any longer because you have now been immunized whether from active infection or from vaccine. I think --
DR. NUZUM: Or both.
DR. FRIEDLANDER: Or both. Exactly. That is exactly right because we think that is what happens.
DR. NUZUM: If it's truly post-event, and that is what I was trying to get to and I got off track on post-event, post-exposure. If it's truly post-exposure, your vaccine is actually boosting infection. If it's post-event but not actual exposure, then there is no danger anyway.
DR. FRIEDLANDER: You are absolutely right.
DR. NUZUM: You won't know.
DR. FRIEDLANDER: But whether you develop an immune response, we know that is a function of the challenge dose when you are on antibiotics. Some animals with a high challenge dose on antibiotics who are protected develop an immune response.
Others, when the dose is low or the duration is long enough but the challenge dose is low, the antibiotics completely suppress the immune response and that was the genesis for the whole argument for vaccination. So I think this is the right way.
I would point out, as I mentioned before, concurrently with this there was a preliminary experiment that was done in the nonhuman primate, as you know, that gives credence to the idea that vaccination coupled with antibiotics would enhance protection.
DR. HEWLETT: Aren't there three variables here for the vaccine? There is the rate of onset, the magnitude, and the duration. The idea here with the post-exposure is to have the rate of onset faster to get to the same magnitude and you don't care about -- maybe you don't care about duration.
But it may be that you could have the same regimen that would accomplish both of those. Maybe you all know that already, but if you could accomplish that and had long enough duration to be useful, then maybe you only need one regimen.
DR. NUZUM: Again, the regimen will be dependent on clinical studies.
DR. FRIEDLANDER: And it depends on the model. In the nonhuman primate we know that the current regimen, at least with 14 days of antibiotics, was sufficient. Now that needs to be expanded on.
DR. NUZUM: Okay. Quickly.
DR. NASS: I think another difference between the pre and post-exposure models is that when you are vaccinating people pre-exposure you are going to have antibody levels going up rapidly after their doses and then falling off over a period of time, and they may be exposed at any time during that fall-off period.
And so you generally will not expect the vaccine to be 100 percent efficacious at any point in time later or until the next booster dose. However, if you are using it post-exposure and you are talking about taking people off antibiotics because you vaccinated them and you think they may have spores in their lungs that have not yet germinated and that they may or may not have an adequate endogenous immune system to fight off those spores like the 88 year-old-woman in Connecticut who probably got just a couple of spores but succumbed none the less because she didn't have much of an immune system.
You are required to have a vaccine that you will get high titers because you will be at the beginning of the immunization period, but you are looking for 100 percent protection if you are going to pull people off antibiotics. I think your expectation of the vaccine is much higher for post-exposure.
DR. KAMMANADIMINITI: Srinivas from Cangene. I have one general question. I understand that in order to be able to demonstrate added benefit of vaccine you are not giving antibiotic treatment to its full potential.
DR. NUZUM: Right.
DR. KAMMANADIMINITI: My question is is this approach acceptable to the regulatory agencies for licensure of vaccine for this indication?
DR. NUZUM: I think what I can say on that is these models have been developed with active interagency participation. These aren't done in isolation but what will ultimately be acceptable to FDA is the data that comes to them.
The reason we are doing this is it provides the public to see what we are doing. FDA gets feedback and really a large purpose of this meeting is to get public feedback that helps FDA determine if this is satisfactory what we are doing.
DR. KAMMANADIMINITI: So it is not yet clear if it is acceptable or not?
DR. NUZUM: It's clear to me that it's not unacceptable.
DR. KAMMANADIMINITI: Thanks.
DR. HEWLETT: When you say not using the antibiotic to its full potential, are you talking about just continuing it longer? I think we are here again at the intersection between the practical and the ideal, the theoretical, because people won't take the antibiotic for prolonged periods. There are complications of doing that. It is the reality of where is that balance point. I don't think it is fair to say you are not using the antibiotic to its full potential because that is difficult to do in reality.
DR. KAMMANADIMINITI: So we can simulate a clinical scenario artificially in the animal model.
DR. HEWLETT: As close as we can get. With the stipulations on the antibiotic pharmacokinetics and the immune response we elicit, you know, that's where we try to bridge as much as possible to humans.
At the end of the day you've got to do -- we've got to do -- I mean, the rabbit -- as Jim Eastep would say if he was here, rabbit is not man and neither is a nonhuman primate. In animals we simulate it the best we can to answer the question we want.
DR. NUZUM: I want to give Conrad credit. I made the comment about the clinical trial to look at vaccine and antibody in combination. This has actually been something that has been discussed for some time. Actually during the break he brought it up so I thought it was a good place to interject that.
Let's make sure this last bullet is covered and then we'll move on to the general wrap-up unless there are other comments on the second bullet. We have already touched on this a little bit regarding what additional data, if any, would be supportive.
There is a comment that to help answer that question we really need to know what the end game is. Certainly that would help. I think it is hard to know what that is. This is the first time we -- as far as I am aware this is the first time vaccines have gone this far using the animal rule.
It's new for all of us including FDA so I think it's very hard to give the end game at this point. I think meetings like this and getting the data out and kind of giving everyone an idea of what we are doing is the best way to help us get to that end game.
I would maybe qualify this question a little and say in the near-term, or in the next steps, or what should we do immediately that we aren't doing to provide data that would be supportive or any ideas on near-term or next steps.
DR. RUBIN: I think one of the first things to do is to try to structure the data that you have in a factorial setting. You have so many factors running around. You have species. You have the vaccine, the dose, the time and duration, antibiotic dose, time and duration, challenge dose.
You get a million factors. Perhaps a dozen factors running around. As you structure the data that has been obtained so far in a factorial setting, you will see that you have some of these cells covered and some of them not covered, I think. I'm just curious.
Has anyone ever thought of trying to design one study that covered several animal species and varied lots of things at the same time and did a fractional replication? Does anyone know what a fractional replication is besides us two?
When you have many factors, let's suppose you have 12 factors like a factors vaccine and each one has two levels, the high/low, antibiotic high/low, animal species, rabbit, nonhuman primate. Another factor, another factor, another factor.
So you have 10 factors, each one at two levels and you have over 1,000 different conditions. Does that say you have to have data in each of those 1,000 conditions in order to understand almost everything that is going on? No.
There is a famous quote that goes back about 75 years to Fisher that says something like, "Nature will reveal her answers most quickly if you ask many questions at the same time." This whole scientific idea of doing one question and then wait for the next one and ask that, that is totally retarded.
There is literally at least a half century of work in industrial experimental design coming to the United States and post-War World II Japan that supports that idea that you don't do it that way. I'm wondering here where we have all these factors where is the experimental design guy or are all these people making up these designs learning it on the fly?
If you don't have this real experimental design guy who knows what fractional replication is and how to analyze fractional replicated data and knows about aliasing and all these other technical terms, you're not doing it right.
DR. SELF: You've got to mention the assumptions underlying this, right?
DR. RUBIN: Yes. The assumptions always are that there are very high order interactions like 10-way interactions are, even if they exist, are un-understandable anyway which is true. There is a 10-way interaction between species and vaccine dose and antibody level, and time -- you never can make use of it in any case. It basically says nature maybe has like three or four-way interactions and that's it. The answer will be revealed most quickly.
DR. KOHBERGER: Don, the issue until this point, and the reason it wasn't done like that was choosing the levels of the factors was unknown. We weren't quite sure what the antibiotic duration should be. We could have chosen 30 days and 20 days. If we had done that the experiment would have been noninformative because they all would have survived.
Until this point I don't think there was sufficient information to know what factors were important and what levels should be important and what the endpoint should be. Now, I think it has come along far enough that thinking about one big grand experiment is possible now where six months to a year ago it wasn't.
DR. SELF: In terms of additional data one thing that occurs to me that would be interesting and useful would be to follow up on this idea of dose and antibiotics, sort of reducing the challenge dose to look at the short regimen, zero/seven, the timing of the challenge dose that might be varied around a range of times reflecting either shorter courses of antibiotics planned or realized by sort of noncompliance of humans to the 60-day course and to look at those times of the challenge at a range of doses that might be commensurate with what you would expect given a course of antibiotics delivered up to that time point.
That might look at the specific regimen that is being described, the short course. It would look at antibodies and drawing those curves in the absence of potential interference for antibiotics so it would be comparable to bridge to some of the GUP data and then could also be used perhaps to compare to those curves that you would draw in the context of the combined regimens. I think that would be an experiment that would maybe link to both sides of this gulf.
DR. NUZUM: Thank you. I have to go back to the pathophysiology, I'm afraid, just for a minute because I've been thinking about the issue of with all the data that have been collected, and I think the studies are really remarkable, I keep wondering if there might be a stratification.
Certainly patients with anthrax have a number of different disease manifestations. Some proportion of them have meningitis, for example. I keep wondering if in the animals that have died have a higher proportion of meningitis than in the whole population.
When we are talking about TNA levels that there might be levels that are related to certain disease manifestations but if your level is here and not here, you are more likely to get meningitis or some other -- liver damage or one of the other manifestations of the illness.
Not to propose more studies, and I'm not sure whether this is in Drisilla's need to know or nice to know, maybe it could go from one to the other, but with the data that have been collected already with the post mortem information and clinical chemistries and liver enzymes I wonder if it would be possible to look at that level of stratification.
DR. NUZUM: I'm trying to think. I think we have done some of that, but I don't know that meningitis is specifically one of the endpoints we've looked at. We do have -- it's a good point. We do have a lot of data like that. We could definitely look at that.
Are there any other comments on this last bullet? Otherwise, we'll go into the general wrap-up.
What I wanted to do here was I have some comments, summary comments, that I've written down, but I also wanted the panelists to each have a last word. The question is do I go first or would you prefer to go first? You want me to go first? Okay.
DR. HEWLETT: You go first.
DR. NUZUM: That's fine. On one hand, I didn't want to steal anybody's thunder but, on the other hand, this way you have the last word.
I think overall one thing we wanted out of this meeting was a sense of whether or not we are on the right track. I think we can say we have heard no major disagreements with the path we're on either for GUP or PEP.
Our basic principle for protective titers for GUP and to show added value of vaccine for PEP seems to be relatively well accepted. I think there was lots of good discussion, questions, comments on what we are doing, and they will provide some good guidance going forward.
The second point I wanted to mention is clearly there is a need for the NHP data, both GUP and PEP. There is also a need for complete characterization of the immune response, cell mediated data, innate immunity, adaptive -- other ways we can characterize adaptive immunity and so forth. Even looking at genetics perhaps.
Actually, some of those efforts are ongoing, and we couldn't present all that here, and that wasn't the purpose today. But certainly to support the animal rule requirements we know we have to look at all of these endpoints to the extent we can.
There was a lot of enthusiasm for passive transfer data and that that work should continue. There was a suggestion to use monoclonal antibodies. I have had an off-line discussion or two that indicates the conclusion may not be universal but certainly among the panelists there was much interest in the passive antibody work.
It may lend another off-line discussion I had on passive work. Even Freyja, I think, mentioned this. The passive work may be a way to provide that randomization of the immune response that was discussed at length.
Another good suggestion I thought, Rick mentioned this, was the idea of a subgroup. I have biologists and statisticians together to not only address this specific anthrax model issue but the animal rule in general, other counter measures and so forth. I think that is a very good suggestion.
There are two groups that don't get together that often. I think if we have heard anything today or in this workshop, there is a need for those groups to be together because this model development to support the animal rule is certainly going to be heavy on statistics and require good statistical analysis and the right statistical analysis.
There was a long discussion about randomization of immune response so that we can evaluate what is going on independent of vaccine dose, double randomization studies and so forth. I don't really know what that means, but I'm sure Bob can help me.
I also heard -- I didn't hear consensus on that point because I think the biologists in the group felt -- there is a sense that if you get similar titers, similar protective titers in multiple studies, multiple species, different labs, different vaccines that is very important. How you get to that titer is less important that you get to that titer by a variety of different means.
It was a very good discussion and I think very thought provoking for all of us. Clearly an area where I think the statisticians and the biologists are a little -- not completely in line.
I think Judy and Jason made the point this morning that best laid plans invariably don't go as you hope. Things take longer than you expect. All of us in science know that is the way it is. We have hard proof here that, yes, it's reality.
Regarding the PEP discussion, I think Bruce made a very good case. A clear-cut correlate will be very difficult to get. I think there is general agreement, or I have a sense there is general agreement amongst the panel that that is the case.
There were some good suggestions on how we might modify our studies. Thinking about what is missing might help guide our studies. Was it fractional --
DR. RUBIN: Fractional replication.
DR. NUZUM: Fractional replication. Okay. I mean, there is a lot here I don't understand statistically. That is the reason for this subgroup of biologists and statisticians getting together. It's too soon to know how that would happen, but I think it is a very good idea.
Okay. Those are my summary points. Maybe we will just start at that end of the table. If you have any wrap-up points, we would be glad to hear them.
DR. SELF: Okay. I think that even though there is a lot of detail being reported in understanding the rabbit model, ultimately the most important information is going to come from meta-analyses across these studies and across species.
I would say that a mechanism to much more rapidly bring data from different studies together, analyze it in a standardized way that doesn't have to be overly complex and present it in that sort of aggregate is probably the most important thing that I see missing today and would probably be the most helpful in planning the next steps forward.
I think the ideas that Don is talking about specifically in terms of being very explicit about what assumptions are being made is going to be useful and I'm not going to say anything more because I wouldn't dare with Don sitting to my left.
I think the GUP studies are on track, but I just have a general sense that there is a little stall because these data haven't been brought together. I suspect there is one that sounds like a very good data set from nonhuman primates.
I worry that once that is lined up with what we know from the rabbits that there is going to be enough of a discord to have to consider seriously some experiments in another species. I know this is a big effort and a big issue but to get that sense sooner rather than later I think would be important.
Then finally for the PEP, as the model is currently configured I think it provides great test of concept information and perhaps some context, but I am worried that the problem is being too narrowly defined. I think it was defined earlier as just what a vaccine can add to a given regimen of antibiotics.
The suggestion at the end of the table to broaden that field and talk about combination regimens and be able to look at both pieces opens up, I think, a whole series of important questions that could be addressed better by modeling and ultimately might have more relevance for humans.
DR. NUZUM: Thank you.
DR. RUBIN: The first thing that I would suggest is to have a glossary of what these three-letter acronyms mean. I would figure it out if it is ever written down just by having one page that I can look them up. I'm serious about that. It would be very, very helpful. It's not a very statistical suggestion, though.
I think the first activity that would be very useful and this is very similar to a comment made, I don't remember by whom but it was already made here, to embed all the current data into one conceptual large factorial experiment where you could weigh out all the factors.
Here are the species, here are types of vaccinations, for example, active, passive, so those are different levels, and where you have data and where you do not have data and what species, and antibiotics given also at the same time situation, challenge doses, and so forth.
It's very hard, at least for me, to keep all the different studies in mind, all the different levels, and all these different things without having them laid out in some organized way. If they are laid out in some organized way, then often you will see that there are certain cells that are missing where you don't have data and why not.
Maybe it was just an oversight. Maybe it was unimportant. Maybe it is unimportant. To keep it all straight I think you really have to get it organized. Always realize that the intention, everybody keeps emphasizing this but I'll emphasize it again, is to extrapolate the results to this one layer of this big factorial table, it's three by five by 10 by whatever it is, to one layer called species that is called human.
The more similarity you see in main effects and interactions and relationships that you can observe in humans and in other species, the more confidence you will have that these extrapolations to humans where you don't have data on humans, where you can't get data on humans will also be similar.
If you find big changes across species and think that it varies even with humans, then that is evidence that you're not there yet. But to the extent that you can get this table of data in such a way that you see understandable relationships, I think that will be very important to do.
I'm not at a level where I have enough feeling for what that big data set with all sorts of missing cells looks like. But always realize you are going to have to make assumptions, and whenever you are making assumptions just be explicit about them because then you can have a debate among scientists. Not amongst statisticians but among scientists who really understand whether those assumptions you are making are plausible.
The final comment is that even though I've been making fairly strong suggestions, I think what has been done so far is really a great job. Very carefully done experiments with lots of thought behind them.
I think in some cases they possibly could have been better with some discussion with statisticians who know about design, but that doesn't mean that I don't appreciate the work that has been done so far. I think it really is a great job. But future work, I think, could be improved if we got together more. Thanks.
DR. GOTSCHLICH: I feel like a lemon that's been squeezed too many times and there's not enough juice left. I think the data and the approach on the general use prophylaxis is obviously on track.
I think very much the suggestion made to provide this in an integrated way so that it is understandable by everybody, even those that are significantly statistically challenged, but I think it's on track. I must say that in the other part I tend to remain stubborn.
I think it's two separate problems that should be dealt with separately. It's antibiotic treatment, and it is prevention of anthrax. I think you will get further by thinking of these as two separate problems. If you want an indication, a different indication, then I think one could think of one that is an accelerated immune response.
Then what you would do is simply design an immunization scheme that achieves that in terms of the protection data available from the general use prophylaxis. I think the experiments were very well done, and they were very worthwhile in exactly pointing out that it's best, in my mind, to separate the two problems. That is my opinion.
DR. FERRIERI: I'll be brief here. I want to thank all of you who organized this. It was wonderful being here and hearing the data. I think what we've heard is extremely positive and gives me a great deal of confidence in how you will move forward and we will be able to accomplish what your original goals were.
I guess my concluding remark really is that all of you who have all the smarts, who have worked in the past, who are currently working on the problem, all the stakeholders in the room need to continue to be together talking through all the issues and sharing your information, or there will be obstacles in finalizing a lot of the goals.
One final comment on PEP. I sort of agree with Emil on that. I'm not sure I quite understand him but there was a dismissal of someone who brought up some antibiotic issues, "That's treatment. We're not going there. We are dealing with prophylaxis."
I would contend that the approach being used if someone comes in post-exposure is treatment and prevention. We need to keep that in mind as we move forward. They are inextricably linked, the issue of the antibiotics as well as augmentation of the immune response. They are not separate.
This is not just all prevention. There is an element here of treatment so if someone comes in, they are going to say, "How are you going to treat me, Doc?" Or someone will say, "What was your treatment?" It's really PEP but linked with that is the notion of treatment that cannot be dismissed and separated.
Again, my thanks to all of you.
DR. NUZUM: Maybe I'll just comment on that. Yes, Pat, I agree completely. It kind of gets back to are you dealing with post-event or post-exposure. I always prefer to say post event because we are not going to know if exposure occurred and MDs are going to use whatever is available.
Depending on when exposure occurred, where the course is they could have the clinical disease that -- well, it wouldn't be clinical yet but bordering on clinical disease and it is treatment. I think that is all going to be teased out between the GUP model, the PEP model, and the treatment model.
I think that is all going to be addressed. For example, I mentioned monoclonals and polyclonals for treatment. They will be looked at for PEP as well. I think that -- maybe we haven't emphasized, but all these years of vaccine model development work is moving or being transitioned into the treatment model work. It's not lost data.
DR. LYONS: Yes. I want to thank everybody, too. I just wanted to reiterate something Art said because the more I think about it, the more I think if there were any more numbers or animals that were going to be used for different things, I think trying to get to a statistical robust measure of the protective level of antibody may be one of the most important things that comes out of at least the GOP studies because I think that is going to come down to being very important for the PEP studies in the future.
If I had more animals put toward anything, I would talking to people like Don and figure out exactly what would satisfy him and Steve and satisfy them.
DR. NUZUM: Is that possible?
DR. LYONS: Not me. I don't speak Don's language I don't think. I think that is where I would focus the initial studies.
Then, secondly, I think Don raised a very good point. I think it is interesting when you listen around the room is that there are -- most people are making assumptions about the models and about what we can know and what we don't know and what we will likely never know including myself.
I don't think we have ever really listed that for anything. That is one of the reasons I think the benefit of having that meeting would be good because I think a lot of people have the same assumptions but let's get them down so that we all understand what the playing field is.
Then, just finally, you know, I really think this is ground-breaking type of work, and I appreciate NIH doing a very good job. To me vaccines and particularly for the next 20 years other than the dominant infections we all know about for emerging infections, though, this may be developing the paradigm for how we do things in the future. I can't tell you how important it is to try and develop a flow chart that we can use. Thanks.
DR. HEWLETT: I certainly have learned a lot through this exercise, and I think just very briefly it is very clear that the vaccine works. It is very clear that above some certain TNA that that's a good indicator of protection.
It does make a difference, however, whether the antibody titers are going down or up at the time. I think that is the difference between the passive and the active.
Bruce and Drisilla both said to me the objective of this is to get scientific buy-in to the concepts that are the basis for this. I think there is no doubt about that. I believe we are working on refinements of that rather than changing the concepts.
Finally, there is not any doubt to the fact that ultimately there has to be a leap of faith. It's a matter of how close we can get, how wide the gorge is but ultimately it has to be that leap of faith and there's not any getting around that.
DR. NUZUM: Thank you. One thing I'm thinking of, coming across the table, several of the recommendations that come out of this could be combined. The data meta- analysis, the table that summarizes all the data, that could be a meeting in itself of the biologist and the statisticians it seems to me. It might be the next meeting whenever that would be.
Are there any comments from the audience? Does anyone out there want a last word? We are just about on time. I'll just wrap up with a couple things. First of all, Sonia Gales isn't here unfortunately. I want to acknowledge her help in starting the planning for this meeting. This has been a very good meeting, and she helped lay the ground work for it.
I especially want to thank Freyja who came in later on here and really has done the heavy lifting. For those of us that had slides to present know that Freyja doesn't let you slip. If you don't do it, you keep getting reminders, but that is a good thing. It's the reason this meeting happened as well as it did.
I want to make a note about the animal studies group and acknowledge them. These studies are the result of years, literally years of, for a lot of it, weekly calls. Now we don't have them as frequent but these calls involve DOD. I'm not going to mention names because I'll forget somebody and there are too many good people. I don't want to leave them out. But, DOD, FDA, NIH, the contractors, you know, Battelle, and the vaccine manufacturers. Depending on the agenda we would have a mix and match kind of call where we may be government only or it may be just the animal contractor Battelle or it may be the vaccine manufacturer.
It's been an ongoing very active, very responsive group, and I think we wouldn't have this data without this group. Sometimes I think how could this model apply to other counter measures and other product lines. I think it is applicable, but it takes a lot of stars to align. It takes time and commitment and responsiveness and engagement from a lot of people.
Then, of course, there is business interest and you get into sensitivity of data very quickly so that really complicates things. Anyway, I think it has been a good model, and we need to think about how it can work for other products. We will certainly keep it going.
Finally, I want to thank the panelists. Their input and experience has been invaluable. I know you are all busy and it's hard to make these trips. I know you travel a lot and none of us like to travel more than we have to. We really appreciate your coming and participating.
That's all I have. Thank you everybody for coming.
(Whereupon, at 11:54 a.m. the meeting was adjourned.)