MR. COMER: Thank you very much, Chairman, and thank you for the opportunity to come and talk about the study that we were asked to do by the Department of Health in the United Kingdom as a result of a recommendation from the United Kingdom's Spongiform Encephalopathy Advisory Committee.

What we were asked to do I think probably was also fairly courageous in the light of Christl Donnelly's talk, which I entirely agree with her conclusions there.

We were asked to assess the magnitude of the risk that could result from the infective agent being present in blood. That's a pretty tall order, really, when we know very little about quite a lot of the factors that could affect that risk, particularly how many people may be incubating the disease.

Nevertheless, being good consultants, we said: Yes, we'll have a go at this and see what useful information can come out from that because we're not just looking at what the actual numbers might be but what actually are the lessons we can learn, what can we actually learn about the processes, particularly what can we learn about which components of blood and blood components are particularly risk factors. Are there particular groups of patients which may be more or less at risk? And can we say anything about the possible effectiveness of the different risk control measures which could be put in place?

Just to look at the time line of the study that we did, the study was initiated following recommendations from the SEAC Committee back at the end of 1997. There was an expert group meeting of a fairly wide range of people in the United Kingdom fairly shortly thereafter.

Our study actually started early in 1998. We did a first draft report in April which then went to review by an expert, group of experts, in the external world, including both members of the United Kingdom SEAC Committee, some of the people around the table here today as well.

Then the final report was produced towards the end of 1998 after a fairly long gap, really, waiting for comments on the revised report. And the final report was then produced early this year.

It is useful to sort of look at that together with the times at which particular decisions were taken in the United Kingdom. In February '98 was when the Committee of Safety in Medicines made initial advice about imported plasma and then the decision, final decision, to implement leukodepletion of fresh blood supply was taken in July 1998, so very much in the process of the time we were working.

SEAC back here in 1997 had advised that the government should consider the use of leukodepletion. And there was a lot of work that was done immediately thereafter.

I think it is also worth just thinking a little bit about some of the reasons for those decisions. Now, I wasn't part of that process, and there may well be others who were more closely involved. But if one actually looks at the press release which the Department of Health issued after that, this is Frank Dobson speaking in the press release, saying that he fully accepts the advice of the Committee of Safety in Medicines. He has decided that the bioproducts laboratory, which is our blood fractionation, plasma fractionation service, will be allowed to import plasma.

And then he says this will reduce the possibility of repeated recalls of blood products in the future and thereby help to maintain public confidence in these products.

So his initial reason was nothing about blood safety. It was about public recall of blood products. And that is reflected very much in the statement from the Committee of Safety in Medicines, from their minutes, where the first recommendation is that a plasma pool subsequently is identified as being strongly suspected of having new variant CJD should be withdrawn -- I'm paraphrasing slightly -- and then to avoid future withdrawals of large batches of medicine or products, including vaccines, manufacturers should avoid the use of U.K. albumin as an excipient to medicinal products, so again concentrating as much, at least, on the risk of recall and the management issues that that arises as well as the health safety implications of variant CJD infectivity in blood.

Just very briefly -- I'm not going to go down these. These were a range of people whom we consulted during the process of the study, including people to do with the blood supply and blood fractionation service for the United Kingdom, people with the Haemophiliac Society in the United Kingdom, uses from haemophiliac centers, so a range of different people, both experts in variant CJD and people involved in the blood business in the United Kingdom.

And then the review panel involved a range of people, both from the United Kingdom SEAC Committee and others, who reviewed our report in detail, came back with comments, which were then taken into account in our final version. So the study has been fairly extensively reviewed and commented.

When we started tackling this, the basic presumption that we had was that variant CJD infections are caused in some way through exposure to the BSE infectivity through the food chain and that will result in a number of cases.

What we needed to do was to then look at what that meant in terms of potential further variant CJD infections through the blood donation route, either through blood components or through plasma pools and plasma derivatives. How many patients were going to be exposed? And what is the potential for an effective unit coming in here, resulting in a new infection of variant CJD?

This is rather similar in a more diagrammatic form of the process which Christl put up, of the way in which you could actually try and model the estimate of infections there from the food supply.

In fact, when we started off, we presumed that in order to get certainly any absolute measure of the risk from the blood supply, we had to try and come up with some estimate of the size or the number of people who would actually be incubating variant CJD.

That was probably the big difference between the early draft of our report and the subsequent draft, when we looked at that issue in more detail and we realized that to try and come up with anything like a best estimate, even with significant ranges, was really not possible, that particularly we know little about the cattle-human species barrier. We know quite a lot about these things pu here, as Christl said. We know the numbers of infected. We know the life expectancy of cattle.

So we know the numbers of advanced infections for the region, but, then, what does that mean in terms of the actual consumption of products and the number of cases which might develop?

So the two big unknowns in there are probably the species barrier between cattle and people and the incubation period for variant CJD when you're crossing a species barrier, in particular.

This slide I won't dwell on. It's, in fact, drawn from the Oxford group's data, again seeing that the peak of infectivity coming in is in 1989. And the bars on here are different ages before infection. Again, I think we're seeing that data already.

When we realized we couldn't come up with any prediction of the number of cases, we decided that the way we would present the risk would be risk of new infection per infected donor. What we tried to do in this slide is just to look at to get some indication of what the potential range might be, which, as we know already, is very large.

What we are seeing here is the fraction of blood donations infected with variant CJD against time and plotted against the mean of the incubation period.

So we've got increasing incubation period up here. And if you see, at low incubation periods, we really have a very small fraction of donations infected: less than one in a million.

As we go out to larger incubation periods, say, if you look at 30, then we're getting up to a maximum of about one in 1,000. They can increase, and obviously they can increase beyond this, too, if one looks at other longer incubation periods. And that's just against one of the potential variable parameters that we have got.

I am just going to go very quickly over the evidence for infectivity in blood. I think probably that will have already been looked at significantly by this Committee, but it was very much part of the background for what we were doing in the study that we did.

If we look at blood transfusions, we know that all attempts to transmit infectivity of blood, blood transfusion, so across a species barrier, have failed and that within animal models, as far as I am aware, the one case which has been reported by Bob Rohwer is still the only case that I have heard of in which there has been a positive transmission by the i/v route within an animal model.

Epidemiology studies have shown that's from sporadic CJD. There is no evidence that there has been any transmission through the blood route. And when we look at blood from human CJD cases, primarily sporadic CJD cases and certainly no variant CJD cases, and look at that, their infectivity through the i/c route into animal models, there have been a few experiments which have shown positive infectivity into rodents but negative results from a significant number of studies into primates and other species.

And there have been some questions asked about -- these cases, these experiments all involve very small numbers of animals and some sort of significant questions asked about those and, in particular, the fact that it is a bit odd that we have got no positive infections in the primates, which you might have expected would be more susceptible than the rodents.

Then when we look at actually within animal models themselves, there have been quite a number of cases, experiments where positive infections have been reported from animals infected with some form of TSE and have been through the i/c route infected in the same species, so again with no species barrier.

So all that we can conclude from that is that the blood from an animal which has been artificially infected with the TSE could contain infectivity. And to some extent, that model may be the one that is most applicable to the situation of people being exposed to a TSE through food exposure.

Again, very briefly, a number of experiments that have been carried out trying to assess what the level of infectivity in whole blood is, ranging here from the low end of about five from some of Diringer's work to over 300 from Casaccia -- again, these are all i/c infective units per milliliter of blood -- and a value of about 10 from the work from Paul Brown and Bob Rohwer.

In deciding what we wanted to use as a base case for the work that we were doing, we decided that it was better to err at the low end. After all, these are all animal models which have been developed to enhance infectivity, enhance the likelihood of infectivity. So when we are looking at the human situation, we would be more likely to be at the low end.

We also have to take into account, as we have already mentioned, that the i/v route, the peripheral route, is going to be less effective than the i/c route. We took a factor of ten for that, again one of the areas where you have got significant uncertainty.

So we took a value of ten i/c infective units per ml as a base case but with a range of values. And we looked at the uncertainty in that and with a factor of ten of the i/v route being less effective than i/c.

We then needed to know what was the level of infectivity in different blood components and in different plasma fractions. The only experiment which has been done which casts any light on that are the experiments which have been done by Paul Brown and Bob Rohwer. Again, I imagine you have already seen a lot of this data.

Two experiments: the spiking experiment, where you have got a high input of spiked hamster adapted scrapie, into human blood, which was then separated and fractionated and all the products of that titrated. I just want to note there, as I know the authors have done, that only a fraction of the infectivity was actually recovered in the final process and that the endogenous experiment, where blood was collected from mice infected with a mouse adapted TSE, again separated and fractionated as before, and then inoculated back into experimental animals.

In the endogenous experiment, there was no transmission for some of the fractions, including whole blood and red cells, but the number of animals inoculated was fairly small. In fact, the expected number of infections for whole blood, for example, would have been less than one.

So what we did was to take the estimate of infectivity in whole blood. I'm now going to talk about intravenous infective units per milliliter. So we've got one i/v, i/v 50 per milliliter blood, so about 450 per conventional units of blood.

We have taken the relative infectivity in plasma and Buffy coat from the Brown and Rohwer experiment, from the endogenous experiment. And we have assumed that no infectivity is lost, so a significant assumption there.

If we do that, we can then get a breakdown of infectivity in the 3 components with about 50 percent of that infectivity being in the plasma, initially a surprising result possibly with the remaining infectivity being about equally divided between red cells and Buffy coat.

Then looking at plasma derivatives, again taking that result for plasma, taking the result from the endogenous experiment, where we could use it for Fractions 1, 2, and 3 together, and cryoprecipitate, and then using the relative infectivity from the spiking experiment for Fractions 4 and 5, we can then get infectivity in the main plasma fractions.

We then wanted to go one step further and look at the infectivity in plasma derivatives, the actual products which were being given to patients.

I have been talking to a number of experts. We felt that there were two alternative ways of calculating that. One was to assume that the infectivity would partition in proportion to the protein content of the product. And the other was to use some kind of estimate of clearance factors from the various processing stages in a blood processing situation.

This slide shows the results of doing that, with the blue bars showing the protein mass content basis and the purple ones showing the estimate based on clearance factors. So this is infectivity assuming that plasma derivative was made 100 percent from infected units. So to get the actual level of infectivity, you then have to multiply that by the proportion of units which were actually infected.

The red line here is unity. So if you're to the right-hand side of that, if you had 100 percent infected blood, then you would have one infected unit per average dose of each of these products. And if you're to the left of it, even with 100 percent of infected blood, you've got less than one infected unit per dose of product.

You can also see that there was wide variation between the two approaches, sometimes about six or seven orders of magnitude here for intravenous IgG, for example, with the protein mass content level giving a reasonably high estimate because you have got high dose about 90 grams, typical dosage for this product for certain patient groups but with a clearance factor basis having a relatively low estimate. So you have got significant variations here.

In the base case results we shall present in a moment, we used the protein mass content basis mainly because they were the more conservative. They gave the higher values. And we used the clearance factor approach as a comparison.

You can see that these two products, in particular, for one type of factor, 8, this is the less pure version of Factor 8. Eight is not much different between the two.

You have got a potential infectivity greater than one. So if you've got high levels of a high proportion of donations infected, you could theoretically get infectivity through this route. And intravenous IgG is the other significant potential.

Here, particularly with this one, this difference is very significant because when we calculated the infectivity for the protein mass content, we took no effect of any subsequent clearance through the processing.

So we were just basing it on the initial infectivity and the protein mass content. And we assumed that subsequent processing steps would have no effect on the infectivity and the product, which is not very likely, I would guess.

What we then needed to do was to look at the way both the blood components and the products are used to actually get an estimate of the risk to the patients being exposed. The way we did that was to define a set of representative patient groups.

There were just not the data available that could have enabled us to look at the way the products were actually used overall in the health service in the United Kingdom.

So, together with medical experts, we defined a set of about 20 different patient groups. We looked at the likely numbers of the patients in each group and the typical dosage to the range of different both blood components and plasma derivatives that they may be exposed to over a treatment period. So these are just some of the patient groups that we identified, and there is more data, obviously, in the report, which you have.

So we defined the treatment and the dose for each of these patient groups, both to blood components and to plasma products. And then by assuming a linear dose response model, we can then estimate the number of new variant CJD infections that could result from that.

And, then, the number of variant CJD cases obviously depends on both the incubation period. And, again, here you're not crossing a species barrier from cattle to people. You're within species. So the incubation period is likely to be less than from cattle to man.

You need to look at the remaining life expectancy of these patients and obviously their probability of surviving the actual episode for which they are being treated.

I'm not going to concentrate on this because I don't think this is the important thing for this. This result shows the numbers of new infections per infected donation for some of the patient groups. So along the bottom here, we have the fraction of donations infected going from unity, on the right-hand side, to one in a million on the left-hand side.

We can see that for many of the patient groups, we're down here at less than ten percent of patients infected for a very wide range of fraction of donations infected.

For some groups, we are at significantly higher level than particularly the patients being given intravenous immunoglobulins, bone marrow failure given red cells and platelets, and acute blood loss being given significant numbers of red cells.

We see this fall off with the fraction of donations infected because with this group, we have a fairly small number of patients. And effectively we have infected all of them by the time we get up to this level. I think all we are saying in this is that there is a range of exposure for different patient groups but highly dependent on the assumptions that we have made.

Overall we estimate that the number of new infections for the base case results are about 2.6 new infections, about equally split between the patients for blood components and the patients for plasma derivatives.

That translates into case of about 0.8. So we've got about 2.6 infections and about 0.8 cases because obviously not all of the patients infected survive long enough to become a case.

Obviously all of those results are highly dependent on the assumptions that we have made. And you can get some interesting insights into that by actually looking at the sensitivity to some of those assumptions.

So here is our base case for looking at new infections, about 0.8 new infections split between blood transfusion cases, plasma derivatives in red, and the green is increased because of patients, recipients continuing to donate.

If we reduce the infectivity by a factor of ten, we see that we make very little difference to the risk from blood transfusion, but we make quite a significant different to the risk from plasma derivatives.

If we reduce it by another factor of ten, we virtually eliminate the risk from plasma derivatives. But, again, the risk from blood transfusion cases stays about the same.

The reason for that is that in a blood transfusion case, you're transfusing typically a unit or more of blood. That unit contains, of the assumptions that we have more, more than 100 infective units of blood. So, even if you reduce it by a factor of 100, you've still got a significant risk of infection; whereas, the plasma derivative results are spread over a very wide number of people with a relatively lower level of exposure.

Conversely, if you increase the infectivity by a factor of ten, you then increase the risk from plasma derivatives very significantly, but, again, you don't do very much to the risk from blood transfusion.

If you look at the incubation period, the base case incubation period for blood supply we assumed was 15 years, so a 15-year incubation period for infection through blood supply. If you reduce that to five, you make a modest increase in the number of cases basically because more patients survive because you've still got the same number of infections but more with a shorter incubation period, a higher proportion of them survive. And, conversely, with a longer incubation period, few of them survive.

So the basic conclusion, the first conclusion, which I think is perhaps important, is that it really is not possible to come up with any reliable estimate of what the real risk of variant CJD infectivity in blood is.

We don't know how many people may be infected, and fundamentally we don't know whether blood from someone with variant CJD could be infective. And we have no evidence to confirm that blood from a person with CJD would be infected. However, evidence with the animal model suggests that there is a potential risk, although we have not demonstrated that that is true yet.

Then looking at the results for the actual study, if there is infectivity in blood at the sort of levels that we have assumed based on the Brown and Rohwer work, then the infectivity that is present in a full unit of red cells would be sufficient to cause infection. That conclusion seems to be valid over really quite a wide range of different assumptions.

Plasma derivatives, the result is slightly different. If we look at the base case and our very conservative assumption that assuming infectivity is based on protein content and taking no account of clearance factors, then there are a few plasma derivatives which could theoretically cause infection. But that conclusion is highly uncertain and varies very significantly over the assumptions that are made, and many of the assumptions tend to reduce the risk, rather than increase it.

So the overall message from that is that looking at risk from blood, it looks as if there's a high risk from the red cell units from the whole blood transfusions than there is from the plasma derivatives. That conclusion seemed to be fairly generally supported by the blood industry people in the United Kingdom.

In the U.K., we have looked at a number of risk reduction measures, including the initial recommendation from SEAC to look at leukodepletion of red cells on the basis that infectivity is perhaps more likely to be associated with white cells, -- that's perhaps a bit uncertain -- eliminate U.K. source plasma, and then a range of other possible measures, including reducing the use of blood obviously would help. Preventing transfusion recipients from giving blood, breaking the recycle loop could be important and possible prophylactic treatment, although there's really no real data on that at the moment.

Just looking at the results of those, again, emphasizing very much looking from our base case, if we look at leukodepletion on that and assuming that the effectiveness of leukodepletion would be to reduce the infectivity by a factor of 100, then we actually see a modest reduction but, actually, a rather small reduction. That may be if leukodepletion is more effective than that or if the level of infectivity in the red cell unit in the first place was significantly less, then the effectiveness of leukodepletion would be significantly greater.

So if we looked at the range of possibilities, leukodepletion could be effective over quite a wide range of different possibilities, but it's not necessarily that effective.

Eliminating U.K. source plasma is obviously a pretty good measure assuming that the source of variant CJD is restricted to the United Kingdom and not from possible source countries, including the U.S. or primarily the U.S., obviously.

So that is very effective in reducing the risk from plasma products, but, as I said, the likelihood is that this risk, the risk from plasma products, is overstated in the study. And it does very little, nothing, in fact, to the risk from blood components.

Reducing the use of blood obviously has an effect in proportion to the amount that you could reduce the usage of blood. There have been some interesting studies in the U.K. where you look at variations between different hospitals in their use of blood for the same operation, and there is huge variation, so obviously a scope there but a sensitive area, I suspect.

Restricting blood recipients from being donators obviously breaks the recycle loop but, again, has some potential implications on the blood supply.

So leukodepletion could have a significant benefit, but the potential effects are uncertain. Eliminating plasma, eliminating U.K. plasma, will eliminate any risk that there is, but the original level of risk might have been extremely small.

And a range of other measures has some possibilities. I think this one received quite a lot of attention in the U.K. recently looking at prophylactic treatment with Pentosan. There seems to be evidence that this could reduce susceptibility in animal models, but there is an awful lot of work to be done I think before we could say with any confidence that that could work for variant CJD.

Thank you.

(Applause.)

CHAIRMAN BROWN: Thank you very much, Dr. Comer.

We have time for a couple of questions. I have a question. I know that a handful of patients who have died with new variant CJD have been identified actually as having donated blood at some point during their incubation period. I know that that ranges from a donation made as early as 1982 to donations that were made just within the past couple of years.

I think -- and this is where I need to be made accurate. I think some, if not all, of those donations were one-to-one blood transfusions or packed cells, but I'm not sure. Can you tell me, for example, if that is true or whether these donations found their way into plasma pools?

MR. COMER: I know for sure they found their way into plasma pools. I do not know the answer to whether they were whole blood donations or not. I think the answer to that is yes, but the policy that they have taken in the U.K. is not to inform recipients, which is a difficult ethical debate, obviously. So I think there has been little publicity about that.

CHAIRMAN BROWN: Right. I know it is wrapped in considerations of confidentiality and patient privacy, but that will obviously be a crucial group to watch and may give you or us the first clue about the reality of whether blood is infectious from patients with new variant CJD.

Of the handful, I think one only or two of the recipients have been alive for more than five years, something like that. I think most of them are just a year or two.

MR. COMER: I think that is right.

CHAIRMAN BROWN: Yes. Questions? Bob?

DR. SCHONBERGER: Could you repeat the answer to the question that you just said? I wasn't sure. It's mostly plasma pools or mostly one to one?

MR. COMER: No. I know for sure that it's plasma pools. I do not know --

DR. SCHONBERGER: It's plasma pools?

MR. COMER: Yes. That is for sure because there were some recalls. I do not know how many were one-to-one blood recipients.

CHAIRMAN BROWN: Bob?

DR. ROHWER: Yes. I wanted to just comment that if I understand you correctly, you are doing your modeling based on the titers that were associated with the crude Cohn fractions in the paper that Paul and I published.

MR. COMER: Yes.

DR. ROHWER: In that regard, virtually none of those materials are used as is. They go through considerable additional refinement before they ever get into people.

We have in the interim completed several spiking-based validation studies, which have some caveats attached to them, of course. Nevertheless, the results have been uniformly very encouraging because we're seeing that in the process of carrying these fractions through scaled-down versions of the manufacturing process, we're seeing the elimination of very high levels of infectivity, suggesting that, at least at the level of plasma fractions, we have another very important additional level of safety that we're getting from the manufacturing process itself.

The other thing I wanted to ask you about was your modeling of the contribution from eliminating donations from persons who had received blood and blood components previously.

I gather you are just looking at the next donation, you are not looking at the issue of propagation of the infection over time by that practice. Is that correct? Because you are showing very little effect here, and in terms of a safety measure, I have always ranked it as one of the most important things we could do.

MR. COMER: That is true. We didn't attempt to model that really fully. And it was just a very crude estimate over the first year. So yes, it is not a full representation of the effect of that.

Just going back to your first point as well, if we take the results from our estimates based on clearance factors, which I think there will be some differences in detail from the results that you have got now with your spiking experiments, if we base the risk from plasma derivatives on the clearance factor approach, then the risk from plasma derivatives is virtually zero. I mean, there really are very, very low levels of risk associated with that. So yes, you get significant, very significant, risk reduction.

CHAIRMAN BROWN: A couple of points just to bring your experimental data up to speed. Unpublished further experiments on the mouse model have produced good news and bad news.

The bad news is that we have a disappointingly large number of transmissions following intravenous inoculation of either plasma or Buffy coat. We also have a transmission using whole blood as a transfusion into these mice. So that's not good news.

The other thing that is not too good is that we have now got in this particular model a ratio of five to one, as opposed to ten to one, which was also disappointing.

The only piece of good news in that in terms of experimental data is that we found that, again, in this model, the level of infectivity during the entire incubation period is almost negligible compared to the level of infectivity during the clinical phase of illness. And that is very good news indeed. So these are data that are not yet published but --

MR. COMER: Can I just clarify that?

CHAIRMAN BROWN: Sure.

MR. COMER: It's five to one between i/v and --

CHAIRMAN BROWN: Yes, i/v and i/c. I mean, we were hoping for at least ten, but that's not the way it happened. Again, there probably is variability from experiment to experiment. And the next time we do it, it might be 10 or 20 or 3. I don't know, but that's the initial number.

Other questions? Yes?

MR. COMER: Well, just commenting on your last point there about the infectivity through the incubation period, our assumption was that levels of infectivity are basically uniform throughout the incubation period, which is obviously the most conservative assumption you could make.

CHAIRMAN BROWN: Right, right. And, as I say, if it turns out to be the case with the human disease, -- and I'm guessing it probably will be -- with you, I think the likelihood of disease, natural disease, whether it be scrapie in sheep, BSE in cattle, or CJD in humans, is going to be quite a lot less virulent than the experimentally induced disease.

Even under the experimental conditions I mentioned, however, infectivity in all components of the blood during the incubation period is so low that it virtually poses I think no risk, at least in terms of plasma derivatives.

Other questions? Yes?

DR. HOLLINGER: Is it your assumption in humans and, say, Dr. Donnelly's in cattle, that all infections lead to cases if followed long enough? That is, is there a chronic carrier assumed to be the case; particularly in cattle, that is? Do we know that at all?

MR. COMER: We assume that any animal infected will result in a case if it survives long enough. That is certainly the assumption I think both of us have made.

DR. HOLLINGER: Is there any data following for prolonged periods of time infected animals?

CHAIRMAN BROWN: There is if -- go ahead. I'm sorry.

DR. DONNELLY: Yes. I mean, I made the assumption, like Philip's group, that all animals that were infected would if followed for long enough lead to disease.

The possibility of carriers, we looked into the possibility of different susceptibility classes. Certainly I don't know of any study that has followed them long enough to be able to -- you tend to have them followed for up to seven years. I don't know of any studies that you do where they're followed for longer to look for these.

CHAIRMAN BROWN: The only study that I'm aware of that documents a carrier state is work in rodents in which mice were treated with Substance X. A few mice that were treated with -- it's the Pentosan-type drug I believe were shown -- maybe they weren't even shown to have infection. They died a natural life without developing clinical disease.

Bob, can you correct me or verify this? I'm not aware now that I think of it again of any study in which infection; for example, documentation by Western Blot or immunostaining of the resistant form of prp, where an animal has carried that all of his life and died from an abscess three years later, which would be the carrier state.

DR. ROHWER: Well, there is a recent report from Rocky Mountain Lab showing a situation just like that, where the animal survived its life span without showing disease, but it could be transmitted, then, subsequently.

There are also some very old papers from Alan Dickinson and his colleagues showing the same thing using certain strains of mice and also depending upon the route by which the animal is infected.

I would just like to caution in terms of thinking about preclinical infection, I think from my perspective, anyway, route and dose could have a very big effect on exactly what we see in these models.

So to date, we have only really looked at the i/c model. I think it behooves us to look at more natural routes of infection before we draw any conclusions about the preclinical state.

DR. EWENSTEIN: I just wanted to make a comment about the use of the plasma derivatives. You have assumed 2,000 units as a single inoculum, I think. I just wanted to make the point that for most patients, there are periods of time when they might receive at least ten times that sort of dose in a matter of days.

Now, I don't know what the cumulative effect is over the space of a couple of days. Over the course of a year, a typical number might be 80,000 units. Again, we don't know the cumulative dose because we don't know the body's ability to clear whatever the infectious agents are.

At least in clinical practice, there would probably be many instances where there would be at least 10 times that exposure in a matter of 48 or 72 hours.

MR. COMER: Yes, obviously what we've done here in looking at the typical -- you know, defining the patient groups and the exposure is just to give some estimates against which we can base some calculations. And there are a whole range of different variabilities that we could look at.

When we actually looked at the effect of changing some of those assumptions, their effect on the results were mainly fairly marginal. So you wouldn't get a big difference by making that sort of a change.

CHAIRMAN BROWN: We have time for two more questions.

Yes, Dr. Leitman.

DR. LEITMAN: This is for Dr. Donnelly. One of the most compelling pieces of data that there's blood transmission of the agent is through the maternal to fetal transmission in cattle, and you quoted a risk of 10 percent over the last six months of gestation.

That's all from clinically observed information? There's no experimental data on that? That's question number one.

And question number two: Couldn't that not also be due to an increased genetic susceptibility to infection in the same -- passed on from the mother to the calf?

DR. DONNELLY: Well, we looked at two main sources of data in looking at maternal transmission. There was the maternal cohort study which was organized by Ministry of Agriculture staff. And unfortunately, rather than recruiting calves just as they were born, they were actually recruited after they had been in farms for a period of time.

There was a maternally exposed animal and a control animal. About 300 of them were recruited. But unfortunately, those animals both in the maternally exposed and control would have been potentially exposed to infectious feed while they were on the farm.

Now, from that experiment alone, it is quite difficult to distinguish whether or not it's maternal transmission or whether or not it's genetic predisposition. And that's because all the experiment -- or all of the maternally exposed animals were recruited as the last calf, so you didn't have a long period of time, a spectrum over the maternal incubation period.

But, looking at the main database, which has been collected on all BSE confirmed cases in Great Britain, we were able to look at those for whom the mothers had been identified and look at dam calf pairs of BSE cases.

And if you do that, taking into account survival of both dam and calf, you're able to see an increased risk for those animals born at the end of the maternal incubation period, but no increased risk for those born two or three years prior to onset.

So that definitely suggests that it is maternal transmission rather than a genetic predisposition. And that, I suppose, is something to note as well in the potential for carrier animals is that genetic studies that have been done have -- with one exception, which was not followed up with additional experiments, have generally not shown a genetic link in cattle and predisposition.

CHAIRMAN BROWN: Is this directed to -- yeah, okay.

DR. PRUSINER: I would just like to ask you one question. What do you think the mechanism is for a cow near the end of its incubation time so it now has high titers in its brain and it's more likely to infect a calf that's born to it than earlier on?

That's what you're saying, correct?

DR. DONNELLY: Yes.

DR. PRUSINER: That's the strongest data you have. The first piece of data that you -- I don't mean to be tough about this, but I think the first piece of data you quote, the cohort study, tells us nothing.

It's zero because of the way the animals were ascertained, they way they were taken into the study. So I think to quote the study constantly is really a mistake. It doesn't -- it's not a clear study. And I think that people in Britain are equally divided amongst what this study means.

So the second study is the one you're quoting now. It's your study. And I don't understand the mechanism.

DR. DONNELLY: I don't understand the mechanism either. I mean, what we were looking at was increased risk as it was associated with incubation stage. And as an epidemiologist and statistician, I don't think we'll ever get at the mechanism in that manner.

One thing that was interesting was an examination of beef suckler calves that John Wilesmith looked at, was to try and look to see what the transmission rate is there. And it was kind of a smallish sample size, but it didn't show any increased risk in those animals that had suckled for approximately a year.

So that suggests it probably wasn't milk because, had it been milk, you would have seen a differential in risk. But otherwise, I don't think that all the statistics in the world and the biggest sample size we'd ever actually be able to tell the mechanism.

CHAIRMAN BROWN: Yes, Linda.

DR. DETWILER: Looking at the database and looking at the calf sample, did you look over the entire course of the epidemic or was it concentrated to a certain point of time with the calves?

Because that might -- exposure to feed, too, during their life span might play a difference in the --

DR. DONNELLY: The data was mainly on BABs, or born after the ban, cases. But we did control for what the risk from feed would have been in their herd. So there was a control for what they probably would have gotten to see the expected number of pairs we would have seen.

So we look at the number of cows and the number of offspring that were cases and how many -- within that herd, how many pairs you would expect. So it is controlled for what you'd expect their feed risk was.

DR. DETWILER: What year specifically, do you have that?

DR. DONNELLY: Oh, born after the ban calves, those would have been -- they were mainly born in the second half of '88, '89 and some in '90.

CHAIRMAN BROWN: Mike, sorry to keep you standing so long. You have a comment?

DR. BUSCH: Thank you. Yeah, just a comment/question.

The hemophilic community often frame themselves as the canaries in the mine, and I think here obviously the British population are the canaries vis--vis transfusion transmission potential. We're ten years out from the peak of the BSE epidemic, and I'm just curious, from your models, at what point in time downstream would you begin to conclude that transfusion transmission is not an issue?

As this committee begins to deliberate, I think it's important to consider any ban that might be implemented on U.S. travel to Britain. How long will that be in place, and can the experience in Britain give us some sense of when we could discontinue such a ban were one introduced?

MR. COMER: I don't think we can really answer that at all because we still know very little about the incubation periods both from cattle into man, so when might the peak of variant CJD cases be in the United Kingdom, and also what the incubation period within the blood supply would be.

We simply don't know the answer to either of those questions. And I think we'll be a number of years yet before we can really use the data to give us a better feel for what those numbers are likely to be. So it's not going to be short.

CHAIRMAN BROWN: Larry, the last comment now.

DR. SCHONBERGER: This would be for Donnelly as well. My understanding is that the oldest new variant case of CJD is in the early '50s. You mentioned that you had data that cattle at different ages had a different susceptibility to BSE.

And I was wondering how strong that data is. You talked about an increase susceptibility between the ages of six months and 18 months, but that the exposures, you implied, were as great under six months and over 18 months as during that period, and yet your statistics didn't show that the cattle were coming down.

Is that what you were trying to say ?

DR. DONNELLY: Well, through the statistics alone of the back calculation, you can only get what's the convolution or the combination of exposure to susceptibility together. But it's by additional data from looking at farmers and what they say they do in practice that exposure seems to be within one order of magnitude about the same all the way through.

But you do seem to have this window.

DR. SCHONBERGER: You mean after 18 months --

DR. DONNELLY: Yes.

DR. SCHONBERGER: -- exposure was just as great, but your --

DR. DONNELLY: Yes.

DR. SCHONBERGER: -- data does not show that they're coming down with the disease?

DR. DONNELLY: Oh, yes; and if anything, it gets greater at 24 months when the cattle start milking. One thing I didn't have time to get into was the fact in doing our analysis of the variant CJD epidemic, in addition to requiring consistency with the annual incidence of cases, we also require consistency with the age distribution of cases.

And in doing that, we're only able to reproduce the age distribution of the cases observed today if there is some age dependency. That can take the form of an age dependency in the incubation period distribution, or it can take an age dependency in exposure susceptibility.

Now, it's difficult to imagine what the biological mechanism, even if you could work it out in cattle, would necessary apply to humans. But also with humans, you have considerable difficulty of hard to quantify differences in characteristics of dietary choices with age.

But there does appear to be something. We don't yet know what it is. But through time, in the next couple of years, we will hopefully be able to get more data to tell whether or not we can distinguish between it being an age dependent incubation period and age dependent exposure susceptibility.

But in the cattle, it's very clear: you can't get a fit to the data just on the basis of constant susceptibility, or even susceptibility peaking at birth and dropping right off.

CHAIRMAN BROWN: Thank you very much, both Drs. Donnelly and Comer.

It's now high noon. And I had been reading the agenda from a draft and inadvertently left out a presentation by Dr. Stephen Nightingale about the meeting held by the Advisory Committee on Blood Safety and Availability about the reserve capacity of U.S. blood supply.

He will speak next, and he will be followed by Dr. Penny Chan. Both speakers have kindly agreed to limit their presentations to 20 minutes so that we can remain on schedule.