VMAC Meeting - Pathogen Load

January 23, 2002


Procedural Comments

by Aleta Sindelar 4

Welcome/Regulatory History

by Dr. Stephen Sundlof 5

Salmonella Shedding Studies: Regulatory Perspective

by Dr. Jeff Gilbert 6

Questions & Answers 20

Conduct of Salmonella Shedding Studies

by Dr. Tom Shryock 40

Questions & Answers 61

Pathogen Shedding Study Design Considerations

by Dr. Jeff Gray 68

Questions & Answers 82

Exponent Literature Search

by Dr. Mike Goodman 92

Questions & Answers 102

USDA Pathogen Reduction Measures

by Dr. Barbara Masters 108

by Dr. Karen Hulebak 118

Questions & Answers 125

Epidemiological Evidence of Pathogen Load Effects

by Dr. Scott McEwen 133

Questions & Answers 144

Factors Affecting Pathogen Shedding

by Richard Isaacson 146

Questions & Answers 162

Human Food Safety Evaluation of Animal Drugs

by Dr. Mark Robinson 171

Questions & Answers 180


I N D E X (cont'd)

VMAC Meeting - Pathogen Load

January 23, 2002


Summarize and Adjourn Discussion

by Dr. David White 182





































Keynote: --- indicates inaudible in transcript.


Procedural Comments

by Aleta Sindelar

(1:01 p.m.)

MS. SINDELAR: Mr. Chairman, members of the committee, invited guest speakers, FDA staff, and public participants, I would like to welcome you all to the second issue for discussion, Pathogen Load. I am Aleta Sindelar, the exec. sec. for this committee, for those of you who have just recently arrived.

I would like to share with you some information about the background materials available at the back of the room. First, let me say that the meeting has been and will be open in entirety, so all information and materials and presented at this meeting will be shared publicly.

At the back of the room, you will find a spiral bound book containing all of the information that has been provided to the VMAC members in anticipation of this meeting. All comments provided for review to the committee prior to this meeting are also made available.

A new agenda reflecting the speakers for this meeting has also been copied for your review. The Powerpoint slides of the speeches presented today, as well as all of the aforementioned materials, have been transmitted for posting on the CVM website.

The conflict of interest statement has been already made for the record. So, please, at this time, let me turn the meeting over to Dr. Sundlof, the director, Center for Veterinary Medicine. Thanks, Dr. Sundlof.

Welcome/Regulatory History

by Dr. Stephen Sundlof

DR. SUNDLOF: Thank you, Aleta. I just read on the agenda that I am supposed to give some regulatory history. And I just found that out, so I will start. Four-and-a-half billion years ago --


DR. SUNDLOF: Actually, I am really not qualified to do that, but we have with us a number of folks who are very well qualified to talk on the subject. But on the regulatory part of pathogen load studies, we have required pathogen load studies for antimicrobial drugs that are administered sub-therapeutically in feed for a number of years.

Dr. Gilbert is going to talk about the history of that, and just a little bit of insight about what we have learned from those studies. This should help to give folks a little bit of background when we get into the next part of the discussion, which is looking at the potential for using these kinds of studies for therapeutic drugs as well.

So I am going to turn the podium over right now to Dr. Jeff Gilbert.

Salmonella Shedding Studies: Regulatory Perspective

by Dr. Jeff Gilbert

DR. GILBERT: Thank you, Dr. Sundlof. Everybody hear me okay?

I am going to talk a little bit about the, as I titled it here, "558.15 Studies, A Brief History." Before I get going, I wanted to say a word of thanks to my friends on the VMAC Committee internally, David White, Aleta Sindelar, Burt Mitchell, Mark Robinson, Karen Lampey, Charles Easton, and Bill Flynn; also to my colleagues in 157 who helped me with this, Dr. Catherine Well, Dr. Karen Lampey, Steve Yen, and Gary Sherman.

And, last, but not least, a very special thanks goes out to Dr. Jean Cooper. Now, Jean is no longer with us. I mean, she is still here on earth. She is just not with us at CVM. She is at CVRH. And, Jean, before she left, did a very lengthy retrospective analysis of all of the 558.15 studies.

I used to rail on her for carrying around all of these papers in the hall. It always looked like she was going to a construction site because she taped all of these spreadsheets together and had a big overview of the 558.15 studies.

I have had to call on those recently, so now I appreciate them a bit more. But what I am going to give you today is basically excerpted from Jean's retrospective analysis. So I want to make sure she gets due credit.


If we think back to the '60s, one of my favorite decades, that is when the issue of the use of antimicrobials in animals, and their impact in the human community in medicine really came to the forefront.

A lot of meetings and gatherings of scientists occurred and began focusing down trying to come to the salient question which is, "What is the impact of antimicrobial drug use in animals on the potential development of antimicrobial resistant food-borne pathogens and their subsequent transmission to humans as food contaminants?"


Then, at the dawn of the '70s, we really began to get down to taking a hard look at this. And in 1970, we had the Antibiotic and Animal Feed Task Force formed here at FDA. This task force was formed to address the safety and effectiveness issues associated with antibiotics administered in animal feed.


What the task force found was that therapeutic antibiotics used to relieve animal disease were thought to pose a small risk, because they are usually delivered at a high dose for a short period of time in young animals. So they thought that there probably wasn't going to be an impact on resistance or pathogen load with that use.

The benefits to animals were thought to outweigh any potential risks to humans. There was an identifiable need for using antibiotics in animals to relieve pain and suffering. And, in general, we have always felt that animals equal a safer food supply. So the task force sort of reaffirmed that.


What they concluded was that preapproval studies would be needed or are needed to support microbiological safety of antibiotics in food-producing animals intended for sub-therapeutic uses only including growth promotion and feed efficiency, the low level uses.


From that came the codification of Title 21, Code of Federal Regulations, Part 558.15, or affectionately known as 558.15. And, in there, if you read it -- I do not know if you have read it. It is fairly short. It is easy to read.

The back end of it is mostly tables that have to do with interim marking agreements, but the front end is the real meat of the resistance and pathogen load issues. So if you have not read it, please do. It is only a few pages long.

Sponsors of antibiotics are required to submit study results demonstrating that their product does not promote bacterial drug resistance only when their product is intended to be administered for greater than 14 days, and for non-prescription use in food producing animals. That is the first major tenet.


The second, and this is codified in B-1 through B-3, sponsors are required to submit all information to the agency on the impact of their drugs on the salmonella reservoir, a.k.a., the pathogen load in food producing animals by specified dates depending on the drug class.

I am going to go through these dates now. They are all way back in the '70s. And, generally, most of them were hit, but a lot of the stuff trickled in well into the '80s, and we are still looking at it now.


But, by July 19, 1973, sponsors were to have submitted records and reports of completed, ongoing, or planned studies including protocols on tetracyclines, streptomycin, dihydrostreptomycin, penicillin, and the sulfonamides.


That same year, by October 17, 1973, were to be submitted records and reports of completed ongoing or planned studies including protocols on all of the other antibiotics not mentioned in that first slide.


The following year, spring of '74, the deadline was for records and reports of completed ongoing or planned studies including protocols on the nitrofurans -- our old friends, nitrofurans, which are now gone.


Later on, by April 20, 1974, there was a requirement to have submitted data from completed studies on tetracyclines, streptomycin, dihydrostreptomycin, sulfonamides, and penicillin, assessing the effect of the sub-therapeutic use of these drugs on the feed on the salmonella reservoir -- this gets to the pathogen load part -- in the target animal, as compared to that in the non-medicated controls.


Then we had the same sort of requirement addressing the salmonella reservoir for any antibiotic or sulfonamide drug approved for sub-therapeutic use in animal feeds. Those are the ones that were outside the first group.


And the same requirement for the nitrofurans, to look at the salmonella reservoir, and this was by September 5, 1975. So that sort of wrapped up the series of dates for the requirements to be submitted.


So that gets us into the study design that came out of these requirements, what sort of study was going to be done. The studies were designed as a set, and which would include a shedding component, and a resistance component. So we sort of had two studies within one.

The studies included a negative control and a treaty group. And the animals were inoculated with a lab strain of salmonella typhimurium, a test strain that had been encoded with analgesic acid resistant marker. The strain had to be shown to be free of any transferrable resistant elements.


Again, study design continued. Salmonella were enumerated and tested for susceptibility. The E.coli, resident E.coli in the animals were just tested for susceptibility. They were not necessarily enumerated. though I think some people collected that data.

Studies were generally about eight weeks in length, and the test animals were not required to be near the market age or weight. We had the major test animals, be it beef cattle, or poultry to swine, but they were not always in or around market age or weight when the studies were conducted.


Some of the parameters that were looked at in these studies were the drug effect on the pathogen quantity, prevalence and duration of shedding organisms to salmonella. What was the drug effect on the salmonella antibiotic susceptibility? And also, what was the drug effect on the resident E.coli antibiotic susceptibility?

I do not have all of the details right in front of me, but some of these were just testing back against the drug, or several classes of drugs, just sort of a standard antibiogram.


As far as guidance on these studies, if you have not read these, there is our old guideline 18 and 19, the human health safety criteria, and animal health safety criteria. You can find these in our website listed at the bottom. It is www.fda.gov/cvm.

Go into our section on guidance documents, and you will be able to find them there and print them off. They are fairly short to read too. These guidelines were a product of the task force.


Getting back to the studies, what were the integrity measurements we took a look at. Did the product have antibiotic properties? That was one thing that was important. We wanted to make sure it actually -- you know, at those low levels, what short of effect was it having on resident bacteria there.

Was there any cross-contamination in the studies picking up bacteria from outside or from cross pins? Were the animal numbers sufficient? Was enough drug consumed to test the highest proposed dose over the course of the study?

Were there any concomitant therapies given, in conjunction with the main drug being tested? Were there any naturally occurred salmonella present came into the test system?


Was the salmonella marker stable? We generally found that to be true. Could the salmonella receive any resistance factors? Could the salmonella even colonize the animals? Was the appropriate microbiology methodology used for the time that the studies were conducted?

What tissues were examined? Were we just looking at cecal droppings, or scrapes from the intestine, or taking out liver, spleen, those sorts of things? How often were the samples taken? And was the study link adequate?

So these were, more or less, all of the integrity measurements that we looked at.


All right. This is the results slide, if everybody can see. There is a lot of numbers on there, and I think everybody has a corrected one. If not, I have put on the back table back there this most current copy, which basically reflects the final decision made on these studies.

There is another one that has 24 passing. That was sort of straight out of the shoot. That was the first cut of what came through, 24 pass. But when you look at it and go back through -- we went back through and looked at every one of these studies, the final tally, and came up with the new numbers, where a total of 44 studies were conducted over those classes of antibiotics on the left.

That is the macrolides and lincosamides, 9 studies; 13 studies with ionophores; 15 studies with unclassified gram positives; streptogram is one; glycopeptide is two; bambermycin is two; and broad spectrum is two; so that totals up to 44.

You can see within each drug class how many drugs were tested in what animal species. Most did the poultry, swine, and cattle. A couple of them just did maybe the poultry, because that is all they were focusing on for the approval.

You can see in our pass, fail, and reject columns the numbers there. If you can add those up, the pass, fail, and reject, they will add up back to the total. This basically gives you a broad overview.

Now, a lot of the studies that fail on the first pass maybe were accepted later on. There may have been some mitigation, or maybe they ran the study over. I have got some of those numbers as far as a repeat column I think in the hard copy, but I did not put it here.

So some of the studies, even though they failed on that first go round, they may have come back later and either did it again or fixed something in the study. I am going to talk a little bit there in a minute about what was wrong with these studies. That gives you an idea about the number of studies we are talking about over the past 25 years.


Why did we reject these studies? There were a number of reasons that we tallied up. The salmonella or coliform susceptibility results just were not submitted. That is always not good to deal with when that was a major part of your study.

The quality controls within the experiment, within the microbiology were not adequate. Sometimes the shedding was just too long to measure any prevalence or duration. When we gave that high dose of salmonella, it just took off and stayed up there, and by the end of the eight weeks it was still -- you know, you just could not make any determination about it.

Environmental control animals were often contaminated or they were not included. These are some of the things that would cause us to reject the study, and then maybe they came back and repeated it and they got a pass later.


Some of the animals failed to meet the inclusion criteria. Maybe they were not shedding an adequate number at the beginning or too much. The data was too disorganized to interpret. We had a couple of those where we just could not make heads or tails about what came in. We sent it back for people to work on again.

And, finally, there were just too few animals in the study. A lot of the studies were run in what I would consider sort of a facilities confining mode where we can only do ten animals, because we only have the size for that or whatever. Today, we are trying to get that to be a little more realistic and move to locations where you can have adequate animal numbers.


Some other problems were identified following this retrospective analysis when we looked back at all of this after all of these years. The drug spectrum matched the salmonella and E.coli in only 2 out of the 44 study sets. This we thought could be an issue.

There is limited information on susceptibility changes in the naturally occurring flora that were there. We took a look at some of the susceptibility changes in the resident E.coli.

But what about all of the other bacteria were there, and what was their significance to public health at the time? So we really did not get a whole lot of information on those.

Artificially high inocula. If you have to give in the ten to the 9th, or 10th, or 11th, or 12th, you know, salmonella, is that a sort of challenge that you are going to come in contact with out in the field with these animals?

The lab strain of salmonella is really not representative of any salmonella that we might encounter. It is hard to say just how viable and how good the salmonella was as a representative there.

Small numbers of animals tested. We are not talking about a lot of animals. I mean you can almost count them on a couple of hands there on some of these studies. They were done in isolators often, just very, very confined sort of work.

So, to summarize, what I can tell you, if we think back to the data slide, especially the one where I had 24 passing the first time, nine of those failed, and seven of the nine had failed because of pathogen shedding. Rather than the resistance part, it was the pathogen shedding that they could not overcome.

There were problems with the design and interpretation of the studies. Obviously, they were designed a long time ago. Science is science, but there are better ways to do studies. And also, with the interpretation we probably could have figured out what to do with the information a little bit better.

Everything that was done was based on the policy and regulation of the time. Salmonella is still important obviously. I can remember in the early '70s, there were TV commercials about salmonella. They were really harping on it.

But is that still the major bug? Do we need to look at other ones? What is it that we are doing now that we need to worried about, as far as the pathogen load and the resistance?

Finally, overall history I think will be helpful in steering any current and future efforts on this topic. I know studies that we have helped design recently or take a look at, we are trying to focus more on the field situation and give us the answer in the animal right before it is supposed to go to slaughter, so that we get a little bit better indicator of what is going on.

So, with that said, that sort of gives you a brief history of what was going on with 558.15 studies from the beginning up through the '70s and '80s, and really until now. And I will take any questions, answer what I can, at this time, to see if I can answer anything else about the history.

Questions and Answers

DR. HASCHEK-HOCK: Could you go back to the table? I had a couple of questions on that.


DR. HASCHEK-HOCK: Okay. First of all, you said two of the -- only two of the drugs matched the spectrum for salmonella. Which classes were they in, do you know?

DR. GILBERT: I cannot remember offhand exactly which ones would have been --


DR. GILBERT: The ones. It could have been the broad spectrum that had fallen into that. I cannot remember.

DR. HASCHEK-HOCK: Okay. Then of the ones in the previous table that we received prior to the meeting, there were like nine that had failed.


DR. HASCHEK-HOCK: Now that we only have five that failed.


DR. HASCHEK-HOCK: So could you tell us how many of those five were actually because of the pathogen load?

DR. GILBERT: Originally, out of the nine, it was seven out of nine. I am not sure how many out of five. My thought was that it was five out of five had failed on the pathogen load, as I recall from the data. I cannot be 100 percent accurate, but I am thinking that it was all five of those. The other two I think had gotten mitigated some other way, or had another issue.

DR. HOLLAND: What was an acceptable challenge dose?

DR. GILBERT: On the salmonella, we saw different doses. I think what happened a lot of times were that there was like a pilot study. Every time they broke out their isolate, they did a pilot study to see what was going to be like the current infected dose, or would give you the best shedding, and what have you, so it varied. You saw anywhere from 107, on up to 1010, something like that.

DR. LANGSTON: Is it correct that these animals were given the drug right up to the point of slaughter, in other words, there was no withdrawal period where they were subsequently tested?

DR. GILBERT: I cannot say for 100 percent, but I am pretty -- yes, that is generally the case with them. They fed the drug the duration. They did not pull it with five days and take any measurements, per se, for those last five days to see what happened.

DR. GLENN: Jeff.


DR. GLENN: What are the dates for guidance 18 and 19?

DR. GILBERT: You got me there.

DR. GLENN: I looked on it and I did not see it, and I am just wondering.


DR. GLENN: Thank you.

DR. GILBERT: I can look that up for you and get back. Well, we can just pull it off the web page. It should have the date stamp on the very front sheet of that guidance document. It should have the date on it, but I do not have those with me.

DR. GLENN: Okay. And then let me back up and ask some very simple questions. I think I understand what pathogen quantity is in the unit of measure, but please give me a definition of prevalence and shedding in terms of the criteria you are measuring exactly.

DR. GILBERT: We were looking at like the numbers of animals out of -- you know, what percent out of 100 were shedding, and the duration was just how long they would go. A lot of times you would see if the drug had some antibiotic effect, even at its lower use level, they may shed for a few days, and then drop off, and you just would not see anymore.

You know, we see that all over the board with some other drugs where they shed all throughout. Generally, what we saw was, you know, it would drop off and go down. And then, at some point, they would say, we are not going to take data anymore because we cannot collect it again.

We cannot pick up anymore salmonella. We have been doing this now for three weeks and cannot find any, so they would stop the study. That is sort of the duration. And then the prevalence was just the number of animals total shedding.

DR. GLENN: And shedding relates to fecal grab samples, or total fecal collection over a period of time? And how is that modeled? I mean, I do not need all of the details, but just give me a little sense.

DR. GILBERT: Right. There were a variety of tissues collected, also feces collected, grab samples in the bigger animals; with the chickens, it might have been just to scrape the lining, and that sort of stuff.

DR. GLENN: Okay. So there was actually slaughter and scraping the lining and the gut?


DR. GLENN: As well as what is in the tract itself, the fecal mass?


DR. GLENN: Okay. And then, another question, regarding the integrity of the measurements evaluated, were there specifics relative to replication and statistical analysis way back then when these were being done?

DR. GILBERT: They did statistical analysis and they were looking for some log difference, so there were some stats run on this.

DR. GLENN: Okay.

DR. GILBERT: The statistics, I guess, compared to the bigger studies, I do not know what you would get out of having ten animals, you know, and what not, the number of experimental units, and what have you. I guess it is sort of flaky when you get down very small.

DR. GLENN: Okay. So that gets me to replication.


DR. GLENN: And, presumably, we did not run a study if we did not have adequate replication, but you have alluded to the fact that we did so. And that was a point for rejection.

DR. GILBERT: Right. A lot of these you would see maybe individual animals housed, and maybe you would have 10 control and 10 treated.

DR. GLENN: Okay.

DR. GILBERT: So there was some replication, but it was not replicated over locations or over time or anything.

DR. GLENN: Okay. And then the last question I had is, in the table regarding failed studies, is the definition of failed related to one of these procedural things on replication, or is it related to an assessment of pathogen shedding, or one of the criteria that were measured?

DR. GILBERT: Right, failed was failed. It failed as things. And a lot of times what they would do is they would come in with what appeared to be a flawlessly run study, but it just failed. It did not work. So that is what we categorize as failed.

The rejected ones were the ones that we rejected for some of these sort of technicalities, that maybe they were able to overcome later and get back. Sometimes they did not. Sometimes they walked away, or did not want to repeat it, or things just changed over time.

DR. GLENN: Okay, thanks. Thank you.

DR. WAGES: There were no standardized studies that were recommended by CVM for the species indicated? It just seems like -- I mean, well, was there or was there not?

DR. GILBERT: Right. Dennis, I tell you, not being there at the time, but being here now, there is -- I do not know what they did back then. I think probably, based on the guidelines, probably everybody came up with a very similar study.

And the fact that these were -- a lot of these were run at one particular location, I think, between the CRO, and the sponsor, and us, you sort of ran into almost a standardized trial for these things, but there was no, here, follow this protocol and do it.

So every one of them was a little bit different, but they did have a lot of similar characteristics too.

DR. WAGES: Were the salmonella strains used for a particular species standardized?

DR. GILBERT: Standardized?

DR. WAGES: Well, as in the same strain used?

DR. GILBERT: I think probably that knowing a little bit about the situation, I think there were probably two, three, four strains of salmonella. And, again, as I alluded to, and we still see this, you pull them out and maybe run them through in a pilot to see what is going to work the best under that phase of the moon without water on that day or whatever.

And so, they were able to pick out the one strain that they thought was going to give them good results, and they would go with that one. It may change in the next study for the next company, but there were a couple of strains.

So there were, you know, maybe a handful at most they were looking at. It is not like they were pulling these out of the field and generating them every so often.

DR. ANDERSON: I notice on your table up there, your third one now is unclassified gram positive. Did you require them to tell you what class of antibiotic they were using?

DR. GILBERT: I do not have that list in front of me; otherwise I could tell you all of the drugs that were in there, and I do not know what was required at the time. They generally told us up front what they were -- you know, what they thought they were doing.

We did not always make them go after and tell us exact mechanism of action on everything. So sometimes -- you know, at the time, I do not know exactly how they classified them. They just fell into the unclassified category.

DR. PARKHURST: One more time please. What prevalence is a percentage or a number?

DR. GILBERT: Well, I guess it could be both. We were looking at overall how many animals were shedding on the day of the collection. If there were 10 animals in the treatment group, we would get an idea, you know, yesterday it was 1 out of 10; today it is 10 out of 10; tomorrow, 10 out of 10, and we kept tracking that. So, depending, it could be a number or a percentage, I guess, depending on how you calculated it.

DR. PARKHURST: Different studies have different numbers, right?

DR. GILBERT: Yes, different studies have different numbers.

DR. PARKHURST: So, in one case, you could be looking at 1 out of 10, and in another case you could be looking at 1 out of 100?

DR. GILBERT: Sure, sure. Again, these are fairly small numbers. We were not talking thousands of animals, or even hundreds of animals. They were more like into the tens or even single digits in some cases. So we, you know, took a look at that best we could to see is it significant that eight out of eight animals are shedding today on any given day?

DR. PARKHURST: So was it a percentage, or is it a number?

DR. GILBERT: I do not have that exact data in front of me. I do not know how it was reported in all of the studies. I think we got probably a little of both.

On this day, a 100 percent of the animals were shedding. Many animals were there. On this day, maybe only three animals out of five were shedding, so we got three out of five.

DR. PARKHURST: And could you just give me a sense of the sample size on these studies that ranged from say the smallest to the largest?

DR. GILBERT: Sample size?

DR. PARKHURST: Number of animals in the study.

DR. GILBERT: Maybe an entire experiment done with, let's say, ten pigs or ten calves versus a study done with maybe 40 versus 40 chickens, 80 total, something like that. So they were fairly small numbers.

DR. PARKHURST: Thank you.

DR. KOCHEVAR: You have already mentioned that the design was probably pretty similar just because the same maybe groups were running the studies. Were the conditions under which the animals were held and handled pretty much the same as well, in other words, the same time of the year? Or were some of them in the dead of winter, and some of them were -- I mean.

DR. GILBERT: I do not have the exact start and finish dates for all of the studies, but they were probably run throughout the year. Obviously, some of the CROs will ramp up or ramp down depending on when they want to take vacation, or what time of the year is, but they were spread throughout the year. They were not always down in the spring time.

DR. KOCHEVAR: I guess what I am getting at is just were there more stressors for some animals in the study groups versus others?

DR. GILBERT: I think the conditions at the CRO locations are fairly well-maintained.


DR. GILBERT: So they are not going to be out in the freezing cold, and then brought in and out, and in and out like that. So they are fairly well taken care of, and I think the conditions are fairly standard.

DR. LANGSTON: Was the issue of pathogen shedding you mentioned, number of animals that were shedding at a time, did this also include the number of colonies from each of those animals at the time? Was that also considered?

DR. GILBERT: All of that information was collected, coliform units, and all of that is collected, yes, and taken into consideration. I have not looked at any of the data analysis going back to the old -- you know, the actual data analysis to see what they -- you know, but they generally would come in and give us as much information as they could.

And, besides, just the numbers and that, they gotten all the way down to the CF used per gram of feces, and that sort of stuff, so that was reported.

DR. GLENN: Jeff, I have another question.


DR. GLENN: I need to go back to a better understanding for myself on the issue of a failed evaluation for a study. I know that we are measuring pathogen quantity, prevalence, and duration of shedding.


DR. GLENN: The fact that you have a failed interpretation means that you have set parameters for each of those criteria. If they are above the bar, they win; and if they are below, they lose. Is that right?

And then, are those all itemized out and available?

DR. GILBERT: Yes, obviously, the review staff that reviewed them at the time had their bars and knew what they were looking for.

DR. GLENN: Yes, okay.

DR. GILBERT: And when you look at each one of the studies, we had actually tallied it up on the sheet with, you know, pass or fail for those three things, duration, prevalence, and what have you.

So we know that maybe they failed this, failed this, but passed this. And, you know, I do not know what other considerations were taken in, but then some sort of overall conclusion was reached and it either passed or failed.

DR. GLENN: Because it seems on the issue of assessing the salmonella reservoir that we cannot fail something on these three parameters in the absence of a whole lot of other information. So I am just asking that as a general question. And you mentioned other considerations must have been taken in, but it is hard to quantify that.

DR. GILBERT: Yes, when you look at any study, besides just the prevalence, and duration, and quantity, there is going to be a myriad of other things that we take a look at that again may have said, well, okay, they failed this one of the three.

Was there anything that we can say, oh, well obviously they would have failed because this went wrong also? There are a whole lot of other criteria that go down through the review process.

But it does all go back to those three things, and somebody had to make a cut on statistically they were -- you know, little a did not equal little b, these were different. So it was, you know, a plus/minus sort of thing there, and they just checked them off as failed.

DR. GLENN: Okay.

DR. KOCHEVAR: This is probably kind of a dumb question. But was the bar the same? I mean the bar did not move? In other words, the standard for all these studies for pass/fail was held constant in a given species?

I am assuming species may vary, as far as where the bar was set.


DR. KOCHEVAR: Once it was set, that was it.

DR. GILBERT: From my perspective, I think the bar was probably the same. Obviously, different people at different times, as time went by, were judging these. But the basic criteria, you know, in conjunction with statistical analysis, you just had to get down to, you know, it either passed or it failed.

And, again, we try to look at each study to see was there anything in there that was real obvious that we could go back to the company and say, you failed, but did you think about this, all the animals were outside overnight in minus 20 weather, you know, or something like that?

And that may have spurred them to, okay, we are going to do it over again. So there was a lot of mitigation and stuff that went into the review.

DR. HASCHEK-HOCK: So, of the ones that failed, were any of those repeated? I mean, if they failed, they failed completely, whether they repeated or not? Is that --

DR. GILBERT: Yes, if they failed, some of them might have been repeated. I do not know exactly how many out of that were done, but those were the ones that -- that column of failed were the ones that we absolutely judged.

These here reflect, this actually says that five failed, and that was the end of them. Those five did not come back. When it was before, when it had the 24 and the 9, I guess 4 out of the 9 either were repeated or were mitigated, and they made it into the pass column.

DR. HASCHEK-HOCK: So, in theory, then, I mean, it is possible that these five could have come back if they wanted to repeat the studies?


DR. HASCHEK-HOCK: And so, some of the ones that were judged as failed could be repeated and potentially pass?

DR. GILBERT: Yes, I am not sure what the policy on the time was, as far as the sequential experimentation, you know, just keep doing it until you get a pass. I do not think that was allowed.

But there may have been a very obvious reason that they failed, and that we could not help them, you know, as far as a mitigation. They could not come to us and say, well, yes, we know why we failed.

So, they went away, and for whatever reason did not come back. It may have been an economic decision, or, you know, the drug just disappeared.

DR. LANGSTON: I believe I recall from the reading that some went back through and were accepted after lowering the dose or --

DR. GILBERT: Changing the conditions of use, those sorts of things.

DR. LANGSTON: Yes, were any allowed just to repeat numbers and see if it came out the same?

DR. GILBERT: I am sure they were allowed to do that, yes. I am not sure which ones of the ones that passed, again, or went through it a second time. The four that I know that I can attest to, you know, I am not sure exactly what happened with them, but they would have been allowed to try again.

DR. LANGSTON: When it was those same, if it was the same, were the two studies pooled?

DR. GILBERT: That I do not know. I would probably pool them, but I do not know what they did back in the '70s.

DR. WAGES: So we do not know. This is kind of the same question I guess. So, in theory, if I had a product, I could have failed three times, and on the fourth time pass, and be kicked over in the pass category. Is that --

DR. GILBERT: Gee, Dennis, I --

DR. WAGES: You do not know?

DR. GILBERT: You know, anything is possible, but I do not know that that would have been the case. I think -- you know, I do not know what the review atmosphere at the time was, but now we really do not like to enter into sequential experimentation like that. Three strikes you are out I guess.

DR. GLENN: I have another question. I did not pursue this enough. One of the criteria measured is the duration of shedding, so duration relates to time. What is the unit of measure, let's just say, in a fecal grab sample?

Tell me. We are measuring the pathogen? Are we measuring the number per gram of dry matter per minute, or per hour? What is the unit? Tell me what I am measuring.

DR. GILBERT: Well, some of those, they were probably collecting samples like every two or three days over the period of like eight weeks.

DR. GLENN: Right.

DR. GILBERT: So that is what they were basically picking up. So, you know, we had this giant eight week timeframe.

DR. GLENN: Okay.

DR. GILBERT: And, you know, at the very start, you may have high numbers; and then after a few weeks, it trickles down to almost nothing, or they become indistinguishable from controls, and that may run out to the end of the study.

DR. GLENN: Okay.

DR. GILBERT: And that is basically what they were taking a look at.

DR. GLENN: Okay. Now, the physiological stage of growth of this animal impacts digestive passage, and I assume got shedding. And so, I assume that ideally these would all be -- you know, within a species they were all conducted within the same physiologic stage of growth and level of intake of diet.

DR. GILBERT: Well, I think we saw a little bit of everything. I do not think it was extremes where we had 300 pound hogs versus weanlings in two different studies. But, you know, for the time, early mid-'70s,, eight weeks probably was not too far off on broilers was it, Dennis? I do not know. What do you think?

DR. WAGES: No, I mean, if you are looking at --

DR. GILBERT: Yes, so eight weeks probably would have been okay for broilers. And the hogs -- you know, I do not have it, obviously, right in front of me every single study.

But, generally, I am thinking that knowing the CROs and the sources of the animals, they were probably very similar each time as they went by. I do not think, like I said, there were the extremes in having small pigs in one, and huge ones in the other.

DR. KOCHEVAR: Am I correct in understanding that if there was cross-contamination that that was a criteria for exclusion, that that was a problem, so that there really was not an attempt to measure transfer of resistant bacteria from the test population to a bystander population, that was not part of the design?

DR. GILBERT: No, cross-contamination would probably have been a criteria for failure or rejected, you know, depending on what it was.


DR. GILBERT: That was the problem. Yes, we would have told try again.

DR. KOCHEVAR: So you really do not get any insight from these particular studies into how easily that was going to move from?



DR. GILBERT: Yes, these studies were not designed to like follow the salmonella and see where it is going.

Any other questions?

(No response)

DR. GILBERT: All right. I am going to turn it over now to Tom Shryock. He is going to come up and I think give us what an industry perspective, or maybe a further history on this from the other side as it were. So give it to Tom.

DR. LANGSTON: Before we do that, there is something I just wanted to mention that perhaps I should have covered just a tad earlier. It is probably not necessary to mention, given the audience is, I am sure, well-informed, as is the panel.

But, of course, the designation "sub-therapeutic" means different things to different people. And there is I think a perception among many lay people in the press that sub-therapeutic refers only to growth promotion, which I believe I am correct in saying that it also refers to some disease prevention uses, and I simply wanted to point that out. This is not necessarily dealing with just growth promotion.

Conduct of Salmonella Shedding Studies

by Dr. Tom Shryock

DR. SHRYOCK: I would, first of all, like to thank the CVM for the opportunity to be here today. It is not often that one has the opportunity to talk about research that you have been involved with from time-to-time in such an arcane field as salmonella shedding studies.

So I am grateful for the opportunity to share some of my insights and experiences, and also to relate to you that I have worked with Diane Fagerberg at Colorado Animal Research Enterprises where much of this work has been conducted over the course of the years.

As far as what I will be talking about this afternoon, I have put a quick agenda outline here before you just to kind of order through this. And Dr. Gilbert's did an admirable job of reviewing the history of the 558.15, so some of my initial slides may be a bit redundant with his.

Some of the comments you will see, however, are going to vary markedly from his perceptions. So we will have to see how we can best address those as we go through.

But, at any rate, what I want to do is go through the protocol, talk about some of the limitations of the study, and then the lessons learned, which may be then applicable and relevant to the proposed pathogen load studies which is the issue before you today; and, finally, to give you some conclusions.


This is one of the slides that Jeff had already showed to you, but I would like to highlight for you that the FDA Task Force in 1970 was the driver behind these studies, and specifically called for certain kinds of data.

With regard to the salmonella reservoir, prolongation, the carrier state, and the prevalence of R factor containing bacteria, just that R factor is kind of an historical term at this point in time, the resistance plasmic, so things have progressed from that stage. But these were the considerations in that task force report.


This was codified into the 21 CFR 558.15 with the following objectives directed to sponsors conducting studies. Basically, a sponsor had to show two things: (1) that the drug did not adversely impact the quantities, the prevalence, or duration of salmonella shed, and this was in comparison to a baseline non-medicated control group; and then, secondly, that the drug did not increase the salmonella or coliform, meaning E.coli resistance, again, over baseline, to drugs either human or animal medicine.

So, really, for the pathogen load section of the discussion, we are really concerned with just that top bullet point, but I will mention these studies in passing as we go through because they are interrelated.

As we found out, there really was no specific study protocol that was outlined in these regulations. Rather, sponsors were asked to consult with CVM on protocol design.


So, taking these regulations then, and trying to actually implement them, became kind of a work in progress over a period of time. As I alluded to, most, if not all of the studies were done in some way, shape, or form in conjunction with CARE.

There were a few that were not. The sponsors did themselves, and then submitted. So, Dr. Fagerberg is quite the guru of the studies, if you will. Generally, these 558.15 studies started out originally as a single study which could assess all parameters.

But then over time they started to evolve into a salmonella study and a separate coliform study, and that evolution of study continued such that currently -- and I say currently somewhat advisedly, because the last time these studies were actually run to my knowledge was probably a decade or so ago.

But right now we would have a quantitation study, a prevalence and duration study, and then a separate coliform resistance study. So you can begin to see that things are expanding in terms of the numbers of studies that would fall into this 558.15 category.


Let me show you the time line, general protocol sequence here, which may address some of the questions that the committee was posing to Dr. Gilbert. There may be some slight differences between the experiences as I have captured them, and as he has related them, and we will try to reconcile those as best we can.

Again, this is somewhat of historical perspective, if you will. There is not a lot of people that are still around that have the hands-on experience and were there when these things were being done a couple of decades ago.

In general, we would have a situation here where, at least two weeks prior to an oral challenge, we would have cultures, weights, feed intakes, clinical observations being taken.

The animals, be they chicks, be they pigs or calves, when they had an acceptable baseline of coliform resistance, which I will talk about momentarily, the study was begun with a fasting period at day minus one. The fasting period could last from 12 to 24 hours depending on the study.

This was to help them establish the salmonella challenge which would occur at day zero. The animals were allowed ad lib feed for a short period of time, a couple of hours, then fasting was reinstituted and a second oral dose challenge was administered.

It is at this point that the feed was administered, and the 56 day or eight week observation period began. The feed was consistent throughout. It was the same lot of feed. The only difference here would be one had medication incorporated, the other did not.

The challenge dose could either be a high challenge, 1011 total CF used in a quantitation study, or 106 total salmonella in a prevalence duration study.

Again, taking cultures periodically once a week, feed intake, so that you can actually confirm that animals were being satisfactorily medicated. In the clinical observations, necropsy tissues were taken for tissue sequestration analysis. There were some variations as far as the length of time. The one that I am using here is the eight week period.


Some other factors that were very critical to the study design are that the animals that were used were either specific pathogen-free, or, at a minimum, salmonella-free.

What was much more rigorous criteria, however, was the 20 percent or less baseline resistance in E.coli, and there were 12 different antibiotics which were tested by MIC broth dilution to make that determination.

This particular criterion was extremely problematic to find animals that had met those criteria. And, in fact, one of the anecdotal reports initially was that the clinical research organization went to lengths that included hiring a trapper in Florida to acquire feral pigs. So this was an extremely rigorous requirement in some of the early studies.

Dr. Gilbert mentioned the salmonella strain. And, to my knowledge, there was, for the most part, a single strain specific for cattle, a specific strain for swine, and one for poultry.

The strain had to be susceptible to as many antibiotics as possible, because we were looking for resistance transfer into the salmonella, and also into the E.coli. We will see some of that data a little bit later. But this is not a component of pathogen load study, per se. Nevertheless, it is mentioned here.

The salmonella strain, as far as virulence, had to be, well, somewhat of a wimpy strain, I guess you could say, in terms of virulence. You did not want to cause disease in these animals. That was not the objective. Nevertheless, as we will see, some of the doses could actually cause that to happen.

I have listed feed here with three parameters that may seem a little boring, that you need a validated feed assay; you need a specific dose which you are going to use, and that has to come from efficacy studies; and you needed a formulated pre-mix product.

And I will show you why this is so important in just another slide. As far as the study groups, and the numbers of animals that were required, basically, it was about 10 to 12 animals per study group. There were three study groups; the medicated, the environmental control, and the non-medicated group.

The environmental control was placed, physically located between the medicated and the non-medicated groups. These animals themselves in these groups were housed individually, generally, in either isolation, in cages of some sort for poultry, or for swine, in isolation or rooms for the calves.

One reason for that, not only the physical separation, but to minimize coprophagia, so that the animals did not reinfect themselves. The environmental control served as the cross-contamination piece, so that basically you could determine if, in fact, there was some sort of introduction of salmonella from an outside source.

Separate caretakers were used for each one of these groups throughout this study duration, and strict biosecurity was maintained. These studies were done according to good laboratory practices, so there was a very emphasis on data quality.


Let me share with you as far as when these studies were done, and why those assays for feed were so important. This would represent kind of a pipeline diagrammatic for generally any industry product, where you would have a discovery phase, find your new candidate, optimize that with final chemistry selection, formulation, and do some toxicology studies.

Then you would take your molecule out and do clinical studies. You would get your manufacturing or the CM&C package put together, and begin your nonclinical studies at that point as well.

This is what 558.15 studies then basically were a late phase situation. You needed to have a dose. You needed to have an analytical assay and a formulated product in the manufacturing. So, basically, you are very late into a pipeline situation because there is a lot of investment. Timing, of course, is dedicated to getting to this particular point.

Why is that so important? It is important for the following reason shown on the interpretation issues. These studies, the 558.15 studies, were on a pass/fail basis. You could go all the way through your development pipeline to come up against this particular criteria, which, on the basis of a total of 30 animals, could basically stop in its tracks the development of a particular product.

So you are kind of rolling the roulette wheel a bit at this point in time. You do not know for sure whether you are going to pass or not.

Here are some of the criteria, as I was able to obtain them through Diane Fagerberg's experience. And, certainly, the sponsors collectively did not totally have access to these pass/fail criteria. And it is my understanding that Larry Rollins with CVM was very heavily involved in drafting just that.


But it goes on to say that that alone cannot be the basis for determination of a public health hazard. That is why this concept of a biological significance came into play. This was an attempt to understand some of the biological variation that you might see in a small group of ten animals.

So, as best as I understand it at least, these would be the criteria for a pass/fail determination for the quantitation, prevalence, duration sections of the shedding studies. And you can see that there is some wiggle room, some wobble here, in terms of the range that you may come out with, in terms of the data.

In the next couple of slides I will show you, actually have some data that will help you to I think picture what this is really trying to convey for this pass/fail. So this may be one source of complication in terms of why a particular study failed and another one did not.

Now it is not clear, at least to me, whether you had to pass as a sponsor all three of these pieces, or whether two out of three was a majority of two to one, if you will, in the past. That was never really made clear to me. Perhaps, somebody else has information on it.

No matter how you look at this, the bottom line is that there is really no evidence that I am aware of that these criteria can actually be related to the on farm or commercial situation to contaminate meat, or even to human health. These are simply study criteria that were outlined.


So here is the first of two data slides, as far as what actually happened in a particular study outcome. This would have been the quantitation study in the medicated group represented by the very thin dashed line, and non-med in the solid line here.

So you can see that after the two challenged doses, you would have a fairly high degree of shedding, about 105 per gram, that went down fairly quickly, about by day ten or so; and then maintained itself at this red line, which is an arbitrarily drawn line just to help you visualize, and about 102 CFUs per gram is where that stabilized.

And you can see that there is a lot of standard air between the different animals that were in the study. But, by and large, there was no adverse effect on salmonella shedding in medicated relative to the non-medicated controls. So this is the kind of data that would be generated for the quantitation study.


For the prevalence and duration study -- and I recognize this will be a little bit hard to read -- there were, to set the table for you, treatment groups here, the medicated, and then the non-medicated groups with the individual animals, in this case, pigs, that were listed out.

In the post-challenge sampling day, ranging from 2 up through day 56, these zeros were ones indicated a presence or absence of salmonella, and this was done with an enrichment broth. So you either had it or you did not. There were some cases where there was not a sample for some particular reason.

So what you would look for here, in terms of prevalence and duration, is the proportion of days positive. And you can see that that would vary depending upon the animals in the study, and it would vary between the treated and the non-medicated control groups.

The duration shedding simply was how many days did the salmonella continue to be shed before you had a consistent pattern of no recovery. And that varied tremendously again between the animals, even in the treated group.

In this case, it was fairly consistent in the non-medicated group. So these are the kinds of data that were generated for the prevalence and the duration studies for salmonella.


So let me summarize some of the industry experience with these studies. As we know, most of these drugs that were tested were gram positive active. They are all in feed. It was very difficult originally, and probably still would be, to get coliforms at a 20 percent baseline of resistance.

And I used Jeff's original figures here to make my 73 percent calculation. That may now vary a little bit. But, at any rate, the majority of the studies, I guess we could say, passed these studies. But, more importantly, it is a variety of antibiotic classes. It was not that one single class failed. It was that there was a pass in each one.

And if you were to even look at the framework document, categories 1, 2, and 3, there were representatives from each category that were tested and passed: virginiamycin, category one, passed; category two, we would have say like swine that passed; ionophores, category three, those passed.

So there is no particular pattern to what passed and what would fail. The failure of these studies -- this might be something to discuss further. Maybe this is due more to the interpretation than anything.

Those ranges for the different study components, how much of that was borderline versus an outright obvious effect on shedding. I do not know. I do not have the ability to look across studies other than the linco studies.

We did mention that failed studies could be repeated and passed. If a sponsor did want to play again, you could put more money on the table and roll the roulette wheel one more time. It is interesting to note that if you passed in one species, say, chickens, you would usually pass in another species, say, swine.

So that is an interesting concept if you think of what I call study creep here, where we started with that initial one study, that soon became three separate studies; and then for each species. So you potentially could have nine separate studies that you are having to do towards just salmonella shedding component.


As far as the limitations, we are trying to take these model studies and say that there is a relationship to commercial farms, to carcass contamination levels, and even to public health that has not been done to my knowledge. I am not sure how you could do that.

So far as I am aware, there really was only one strain that was tested for the majority of the 558.15 studies. And if you recall that time line, the salmonella was given prior to the medication, which would allow basically the salmonella to slip by the normal flora, if you will. Those animals were fasted.

So the salmonella are already intracellular. They are established by the time the medication is then administered. So that may have some important ramifications later on.

We mentioned high challenge dose. That is not realistic. There is not too many animals that bump into 1011, colony forming units of salmonella. That dose itself can overcome a protected flora. It has caused disease in some of these animals as I understand.

So you not only have a situation that can compromise the study integrity where sick animals do not eat as well, but there is also some potential for welfare concerns as well.


And then trying to take the model study and compare that to the field situation, what are the similarities? What are the differences?

But I think, you know -- I won't go through all of this, but there is a lot in terms of the challenge, the housing, the natural exposure proportion of animals that have seen or not seen salmonella.

I guess the bottom line is that in the model studies, this was a research grade status. Everything was ambient temperature, well-controlled, best of feed that you could get, great biosecurity, on and on, not very reflective of the real world situation.

So the model study really could not factor these particular real world parameters in, because you needed to control them so carefully.


So what lessons can we apply from this 558.15 experience to pathogen load studies? Well, really, all we have got is an in-feed medication database on a multi-week basis with salmonella challenge. That is all we have got to really draw from and try to extrapolate further.

There really is no experience with higher doses, other routes, shorter durations, post-medication withdrawal effects, et cetera. And as you consider therapeutics and how they are administered today, how does that play into this whole situation? When should we sample? Is it even relevant to do these studies?

We learned that there was some study creep that was beginning to enter in. Will this also come back and we will need a separate study for salmonella, one for campylobacter, one for E.coli 0157? How many studies will actually be required?

There is little experience with broad spectrum drugs, say, flora quinoline, or a broad spectrum cephalosporin. Those are likely to actually decrease shedding. What are you going to do with that?

I am not particularly sure how that is going to be address. We also know there is a lot of biological variation. Ten animals is what was being used here. That is a small number.


I would just like to refresh your collective memories here as far as some other events that led up to this particular meeting. There was a preapproval workshop held here at the Double Tree, where these groups found no value to conducting pathogen load studies.

So that is on the record. It is on the CVM website, as far as transcripts and presentations. The exponent report, which we will shortly be hearing about, again, found no consistent evidence to indicate an association of salmonella shedding with antibiotics.

And, again, going back all the way to the 1970 FDA Task Force recommendation where therapeutic drugs were excluded at that point. So right now we are in a situation where there is not a pathogen load shedding protocol, per se.

And it would be a shame to try to go back and modify this on a pay-as-you-go basis, so which we did with the 558.15 studies. And, to be frank, I do not think that is an acceptable way to do business.

I will draw your attention to the fact that already some sponsors, to my knowledge, have been asked by the CVM to do some of these pathogen load studies. Whether those have been done or not, I do not know, but I am aware that some people had been asked to conduct those studies to pursue some of their registration claims.


So the relevance of the pathogen load studies, just a couple of items for your consideration. If you look at the swine population, about 8 percent could be considered salmonella positive, HACCP pork carcass baseline.

We actually get to the meat and it is about an 8 percent positive rate. The feed additive use is 90 percent. That is data from the NAMS. So is there a relationship there? Have these studies done their job? Just something to think about perhaps.

So, in terms of trying then to apply a pathogen load situation to therapeutics, how is that going to really impact the real world?

We also have to consider that if you look at human salmonella serotypes, 6 of the top 10 are different from those that are in food animals. So what strains are you going to use to predict impact on public health? A difficult question in study design.


We also know that salmonella load can be affected by a variety of external factors, not just antibiotics. For example, transport stress, feed, and environment, all sorts of things can cause a shedding non-specifically.

We also know the contamination of meat can occur post-slaughter, but one could assume that it is a farm origin instead. So there are some issues there as far as how does all this apply?

Already there is a number of interventions that are in place, and there are some even on farms, such as vaccination against salmonella, that are being utilized. But HACCP, and cooking, and a variety of other interventions already are serving to minimize food-borne pathogen contamination.

If you think about it really, what the producer is delivering is muscle to the slaughter plant and it is sterile. The muscle would be sterile. It is only upon the cut up that you would have contamination occur on the surface of that meat.

So that is a real important critical control point, which is perhaps outside the purview of a preapproval study. But it is something that does occur, and does go on.


So, to conclude here then, keep in mind that these 558.15 studies are in place now for in-feed antibiotic products. Those have not changed. They are still on the books.

They are still required for response or wanting to take a product through that particular pathway. The new twist to this is that the pathogen load studies would be required for therapeutic antibiotics.

And, to my knowledge, I struggle with having to try to design one of these that it has got relevance to an on farm practice, to meat contamination, and, ultimately, to subsequent human illness. To try to link those is for me a stretch.

And, finally, you know, to be frank about it, I think these are basic and unnecessary impediment to the new animal drug process for therapeutic antimicrobials. The more burdens put on the sponsors along these lines provides additional disincentive to come out with new products.

In terms of safeguarding public health, I think there is sufficient other avenues that are already in place that will provide that assurance. So, with that, I will close. Once again, I will thank CVM for the opportunity to share my thoughts on this. And I will be happy to entertain questions from the panel. Thank you.

Questions and Answers

DR. WAGES: Tom, do you have any information on if this drug sponsor is being requested to actually do some of these pathogen load studies? Are they a specific designed study to be performed?

DR. SHRYOCK: It is a situation where you would need to consult with the specific individuals involved and discuss a protocol design. I really cannot speak for other sponsors as to what they may or may not have agreed or not agreed to do, and just mentioning that there is some discussion ongoing between the agency and a sponsor. So I cannot tell you specifics, Dennis.

DR. KOCHEVAR: Just to clarify the statement in your slide that the preapproved workshop group found no value in conducting pathogen load studies, that was specifically related to therapeutic use, not sub-therapeutic use?

DR. SHRYOCK: That is correct.


DR. WOOD: Have any optimum models been developed for a therapeutic study? I mean you have pointed out problems with the current dealing with sub-therapeutic antibiotics. But have any models been developed that this might be an approach that would work?

DR. SHRYOCK: I am not aware of any models along salmonella shedding lines that have been developed. There may be some research reports that are out there that are semi in that vain.

But to have them with the rigor and the control that meet the 558.15 study criteria, I am personally not aware of those. And I do my utmost to keep current with the literature at this point and on these topics.

So I would be welcome, open to other inputs to change my mind on that, but I am not aware of any such study designs.

DR. GLENN: Tom, I am sorry.


DR. GLENN: I wanted to reiterate something that you said that I think -- I realize we are getting your perspective. But it is somewhat disturbing that a standardized, validated protocol with clear interpretation of results does not exist for pathogen load studies.

That implies that there is no reason to do a pathogen load study, that we are wasting our time. So that is a big mouthful there, and I am just making that comment. We will continue to assess that, and I would hope that the committee, you know, gets enough information to get our arms around that one.

DR. PARKHURST: What is the status of the historical data? Are there raw data that are available?

DR. SHRYOCK: All of the information, as far as I am aware, are proprietary and would reside at some point with the CVM or the sponsors that submitted them. I am aware that there are studies that have been done, but not submitted, because an initial evaluation by the sponsor would have been such that they felt they probably would not have passed an approval situation.

The only other way to find these bits and pieces would be to look in the literature as the exponent report has done. And they certainly did not capture all of the studies, but some people have reported 558.15 studies in the literature.

So there are some examples there. But, certainly, not all of the compounds and studies are out there for ease of access.

DR. ANDERSON: I got the feeling that you think the -- what you said is that you feel like the pathogen loading should be done away with because other avenues already in place to safeguard public health are available. Can you describe those avenues?

DR. SHRYOCK: Certainly. For each and every pathogen that you can consider, there is probably a specific intervention on farm that could be considered for a given species group. So that would take up more time than I think is appropriate.

But there is on farm activities as a start that could be implemented, adequate nutrition, housing management, vaccination, et cetera. So I will leave it as a general on farm there.

When you actually bring these animals to a slaughter or processing facility, again, there are interventions in place there that we collectively rely upon such as HACCP and other mechanisms to assure that the contamination is held to a minimum.

And products do make it to the marketplace with salmonella on them. These are baselines. They are not zero levels. So anything that comes out of that plant has met those criteria for HACCP.

A final safeguard as I see it would be basically good kitchen hygiene, cooking, preprocessed meat products, et cetera. So each of these steps is designed in a HACCP mode, if you will, to reduce the risk throughout the entire food chain continuum.

So those are the safeguards that I perceive are in place and are serving us effectively. There is room for improvement, but there always will be. But those are the ones that I would see as key.

DR. KOCHEVAR: I had two questions. Back on about slide 10, you make the statement, "Antibiotic resistance, moot point, no difference is ever observed. Those really were not part of these studies." Is that true?

DR. SHRYOCK: Yes, thank you for pointing that out, because I did gloss over that. In terms of the salmonella that were recovered in the studies here, either in the quantitation or the prevalence duration aspect, those were actually tested against a panel of 12 antibiotics representing different classes of antimicrobials which were important to both animal and human medicine.

The finding was that there was no change in the antibiogram of the isolates relative to the initial challenge strain. So the coliform study would have been additional reinforcement of that.

And my understanding from Diane Fagerberg, at least, is that she had not seen any changes across the board for all of the sponsors products that she had tested. So, basically, there was never an issue, as far as I know, when the CVM reviewed that that a product was failed because of antibiotic resistance.

Again, keep in mind that part of the reason for that as well that the products tested are gram positive active. Salmonella, being gram negative, you would not expect it to do anything. So, from that sense, it is good that nothing happened if you were to expect that.

DR. WOOD: I just want to make sure though that you were not implying, were you, that on farm intervention is -- I see it the other way. We have always understood that on farm intervention is very important in this whole process, and also in terms of reducing pathogen load.

A good case and point was in the supreme beef, you know, it was determined that the problem could be traced back to the farm in terms of the salmonella pathogen load at that point; and, certainly, in other studies, like by Robert Tuckson, and some European studies as well that sees the connection.

So is your premise to question the need for on farm controls, or is it to question the appropriateness of there being therapeutic studies for pathogen load?

DR. SHRYOCK: I, in no way, meant to imply or denigrate on farm controls and interventions. I think those are very important, and they serve a very critical role.

Where I am coming from is that in the context of the drug approval process, pathogen load studies are done in a preapproval mode situation, in a very artificially done model situation that would be very difficult to use the data to predict an affect, either in a commercial situation, a commercial production situation, meat contamination situation, and then ultimately to a potential impact on public health disease incidence.

That is the disconnected that I find difficulty with. Does that help?

DR. WOOD: Yes, it does, thank you. Is there any post-approval role for measuring pathogen load?

DR. SHRYOCK: There are no requirements to do so that I am aware of. Many research investigators at academic institutions might well choose to look at that. I know that many production and producers would consider those kinds of things if they do monitor salmonella in their plants. And if they saw a blip, they would ask the question why, and try to look at it from that direction.

DR. KOCHEVAR: I guess I am still worried about the antibiotic resistance part of this.

If, basically, these studies contribute nothing to our understanding of whether or not antibiotic resistance occurred because the antibiotics were not matched to the bugs that were looked at, I mean, you would agree in these studies?

DR. SHRYOCK: That would be an accurate assessment, yes.

DR. KOCHEVAR: And so, in the workshop on therapeutic use of pathogen load, we really do not have any data to evaluate that, since the data from these studies is no use for that.

DR. SHRYOCK: No, the preapproval workshop study discussions. And I only was in one of the four species groups, so I cannot speak to the others. But my perspective in understanding just when we had the summary statements was that these kinds of studies in general would not lend a whole lot of value to a drug approval process. And a resistance component is covered yet in another dimension within the framework document. Thank you.

MS. SINDELAR: Our next speaker is Jeff Gray.

Pathogen Shedding Study Design Considerations

by Dr. Jeff Gray

DR. GRAY: Committee members and meeting attendees, I appreciate the opportunity to discuss some things about pathogen study design considerations. I think when you are discussing salmonella and pathogen load, there are volumes of data out there to consider.

I think one thing from discussing briefly with the speakers ahead of this meeting, you will see some overlap between the talks of areas that we all think are important issues for you to consider. And you may well want to pay attention to those.


But I am going to discuss study design considerations, and I have the talk divided up into four areas: Organism characteristics, where I will discuss a little bit about host range and clinical status of the host animal; study design, measuring the effects, and confounding factors.


The first slide I want to discuss a little bit is this one in which we list. I have the top isolates from a given year, and I believe this is 1995/1996. And these will vary year-to-year, but there are some things here that are useful to point out.

When we consider swine and human isolates of salmonella, you can see that some of the serotypes on the list overlap and some do not. And you also have to consider that when we are considering cattle, turkeys, and chickens that the list on your left-hand side are going to differ a little bit.

The other things that is true about these pathogens is all of the salmonellas on these lists are not created equal. They differ in their ability to invade in the whole species, as well as in humans, from serotype to serotype. And I will discuss that a little bit further as we go on in the talk.


Now salmonella is a broad host range organism in general, and we generally term it as being ubiquitous. We can find it all over the place.

And, oftentimes, if a colleague or a student comes to me and says, well, I have a herd that I believe is negative for salmonella, a convention production herd, I tell them you have not looked hard enough because salmonella is there. It is very ubiquitous in the environment.

And the next point that clinical status is questionable. In animals, especially, as I noted in that first slide, we have a broad range of salmonellas that can infect animals. Not all of those salmonellas will cause clinical disease.

In fact, a good majority will not, so there is no way to visually measure whether an animal is sick with salmonella. So they are going to carry that organism whether or not they are sick.

Now, I will mention a few other organisms as we go on in this talk just as a comparison. Campylobacter is much the same as that group of salmonellas in animals. We have a lack of clinical signs in food animals.

We know that campylobacter jejunae in-coli exists in cattle, and swine, and chickens, but we have a lack of clinical signs. E.coli 0157:H7 we know is the same way. We have a much narrower host strain here with cattle being a known host, but we have no clinical signs in food animals.

So these are things to consider when we are trying to figure out what in the field is going to be positive and how to measure what is going on in a specific study for salmonella, or another food-borne pathogen.


Now, something else that is interesting about salmonella is that it can be invasive in the host species. So, in other words, it is not an organism that will stay in the tube, that is, the intestine. It does not necessarily stay in the lumen.

It can invade lymph nodes, and it can go septicemic, in which case, we would see clinical disease. In that case, we are going to see invasion of other tissues. But we clearly know that salmonella likes very much to hang out in lymph node tissues.

And that brings us to the shedding designation. Oftentimes, shedding is viewed as we can take feces out an animal and we can isolate the salmonella. Well, we have carrier animals that are indeed very much positive for salmonella, but in fact are not shedding at that given time.

So, given the right stress, a change in diet, the change in environment, we may in fact induce shedding, because they are positive in their lymph nodes.

The other thing about shedding that we need to take into account is how to measure that. And I will discuss sensitivity of sample taking in a bit. But how we measure shedding, whether it is by fecal swab, whether it is by a sample out of the pen, or whether it is by a large fecal sample from that animal can be very different in the sensitivity of that test.

Now, again, as was mentioned in the previous talk, we have strain differences among salmonella isolates, not only do the serotypes differ between their ability to cause disease, the level that they are going to shed, and how long a duration that they may shed, individual strains within a serotype can differ very greatly in their ability to do this.

So, if we take five salmonella typhimuriums, there can be large differences in what you see in the outcome after you dose a group of animals. I am going to move on here to infectious dose, and some discordance that exists between shedding and infectious dose that we see.


Now, if we look at challenge studies, and we look at challenge dose, and this is an experimental situation where we take and we either give 103 CFU, 106 CFU, or 109 CFU salmonella in swine. With 103, what we typically see is no detectable shedding, and we have seen no deep tissue infection.

So, in other words, when those studies were done, we could not easily go back and isolate the organism. However, if we took 106 CFU, we got a 101 shedding peak, as was measured by the techniques used, and we got carriage in deep tissues out to eight weeks.

If we go on to a 109 -- and this would obviously be a curve of doses, we go on to 109, we have a 103 shedding peak, and we have long-term carriage, well over 12 to 16 week carriage in these animals.

So, now, if we look at controlled natural infections, these are infections done in experimental environment with previously known salmonella negative animals. However, the infection process would be done differently where a seeder animal is infected and put in with a group of negative animals.


In those situations, what we see is a 103 shedding peak of the seeder animals. However, all of the naive animals that came in contact with that seeder animal also had a 103 shedding peak in those "naturally" exposed animals.

So, if you look at that previous dose study, in this one there are some measurements that do not add up, and those are some things that really have yet to figure out whether what we took out of the lab and gave to the animals was indeed not in the right state, the salmonella was in the right state to infect, or whether the salmonella that the level or the sensitivity of the tests were able to measure using the same technique study-to-study really does not give us an accurate measurement of what is coming out of the pig.

But, in fact, if we had -- in this situation, if we had a 103 shedding peak, and we are assuming that all of these animals were exposed and had a long-term carrier state, the level of coprophagia that would have had to occur there are indeed thousands of grams of feces, and that probably is not occurring.

So there is some questions there that in the literature have not yet been answered.

Now, if we also look to sort of further on that point, swine infected with infected desiccated feces. And what this is is taking a controlled group of animals, infecting them with salmonella, collecting the feces, and allowing it to dry out for a three month period.

Then we go back and we infect new animals with that feces. We can again go into a state where all of the naive animals were infected, a 105 dose was given. There was indeed no shedding.

But if we look at the deep tissue infections in these experiments, we had a 103 level of salmonella in the liver, the spleen, and the ileocolic lymph node. So, again, there are some questions there that with infection dose and shedding that are difficult to add up.


Now, if we go on to look at measuring of the effect of your experiment and maybe a treatment that you are given, you have to consider the sensitivity of the measurement you are using on the back end.

If we looked in 1998 and 1999, some of the papers that were put out, we believe that E.coli 0157:H7, prevalence and feedlots was 5 to 8 percent based on the techniques that were used.

In 2000 and 2001, we now believe that the prevalence is 23 to 25 percent. The only difference in those studies was the technique used to detect E.coli 0157:H7. So it does not mean either study was wrong. It means that the sensitivity of the methods changed over time, and therefore the bar changed over time.

And so, here differences are based on detection methods. So when you are looking at a study you have to ask, are the detection methods used sensitive enough to detect treatment effects? And there needs to be a measurement of that when you are looking at a study. That has to be demonstrated.

Continuing on, if we look at temporal measurements of salmonella shedding, if we have a salmonella high dose challenge, and we look at the shedding curve -- and Dr. Shryock showed you some data like this -- if we look at CFU per gram of feces, we have a fairly predictable shedding curve that goes from day 1, on this slide, goes out to week 7.

And, as you can see, it goes up very rapidly, and then declines slowly over time. And at what point do we want to intervene on this shedding curve? At what point is a treatment given?

We have to consider all of those issues because where we intervene is indeed going to effect the outcome in this shedding curve. And I think that is yet another point you need to consider is that all is not created equal when we look at the shedding over time of salmonella.


If we look at times of intervention, if we intervene prior to infections, what we would probably have to assume here is that we are altering the flora in the intestine so that we are either taking away binding sites, or we are promoting the growth of bacteria that are harmful to the salmonella.

If we look at if we intervene during peak shedding, we have to decide when is that peak shedding. Are we basing it on clinical signs? Because in the field we certainly cannot do that.

And we have to then take into account are we going to cause the induction of a carrier state if we intervene during peak shedding or late in the production cycle. Are we going to drive that salmonella into the lymph node and create a carrier animal?


So, again, what measurements are used, method sensitivity? Are we looking at feces? Are we looking at feces off the floor? Are we taking an adequate sample out of an animal?

If we are taking swabs out of the rectum of an animal, there is data out there that indicates the sensitivity of that, and that is not -- the sensitivity of those methods really are not really very good.

So at what time point are we measuring that? For how long are we measuring the shedding? The idea of no salmonella in the field is not currently feasible, okay. We have to assume that salmonella is going to be there. If salmonella is in many animals, we have to assume it is clinical.

Normally, what we see is it is detectable in a few animals at low levels. So is that the state that we should be looking at when we consider pathogen load and its effect detectable in a few animals at low levels? And that is a hard situation to reproduce. And then what is the food safety risk in those situations?


Now, seasonal changes, this is also important. Winter seasonal prevalence of E.coli 0157:H7, we know to be 3 to 5 percent; summer, 23 percent. Salmonella, we have species and serotype differences to consider, and we have a peak season prevalence of 8 to 12 percent.

That does not lead very much leeway to create a real -- to look at your method sensitivity, and look at what effect you are having on pathogen load. When you are only considering an 8 percent prevalence, you are in your peak season.

Diet can be a confounding factor. Composition of feed stuffs can effect prevalence in transmission of food-borne zoonotic pathogens. So you have to consider when you are looking at a study what was the diet used? Is that reproducible? And is it useful to use that diet in a study?


Stress on the animals. The first thing you have to consider is the manner in which stress is measured. We know that we can have an effect on pathogen shedding with stress. However, stress in a lot of these studies has been used in a very general way.

And if you look at a classical, which has often been used with salmonella, put an animal on a truck and ship it, which is what happens to all of the animals, but that does not create a reproducible stress to cause predictable shedding.

And I think that is going to be touched on a little bit later, so I won't go any further. But Dr. Isaacson, as well as a few others, have done those studies, and that unpredictable stress, while it can have an effect on pathogen shedding, it is hard to predict what effect.

Lairage exposure. Recent swine data indicate that lairage may be an important exposure point. In other words, what has been recently shown by Scott Hurd and his colleagues is that when animals -- swine are held in lairage.

The number of animals that are infected at that point before lairage is much lower than those during lairage and post-lairage directly into the packing plant. And, therefore, that late in the process, we need to consider how are we going to effect pathogen load at that point.

Deep tissue infections. Again, we have salmonella serotype differences here, and just to reiterate that point. Some very much like to be in the lymph node, and we need to consider that. And we are looking at the actual salmonella used.


Environment. Treatment effects may be environment-specific. We know that to be true. The mechanism of action of treatment, remember the temporal relationship of the shedding curve? Are we having an effect on flora or directly on the pathogen? And that is going to effect our outcome. So you need to consider the mechanism of action of the drug.

Long-term survival and infectivity of salmonella. We know that salmonella survives very well in a desiccated state, and it will survive in the environment long-term. So when you consider pathogen load, that can indeed be an important issue, how much salmonella are we putting into the environment?


So to wrap things up here, organism characteristics. Studies must adequately account for specific characteristics of the food-borne zoonotic pathogen. One thing to consider is while studies probably cannot do 25 different isolates within a single serotype, we need to have a measurement.

A real time today measurement is, did the isolate used carry the appropriate virulence factors? When we looked at it, can we do the molecular analysis of that isolate? And is it normal compared to what we expect to see in the field?

Study design. We need to model a realistic infection cycle and a target host species.

Measuring the effects. The measurements must be robust enough to show differences under realistic circumstances.

And confounding factors. They must reasonably account for factors in production systems and pathogen strain differences.

And, with that, I will give you a view of the challenges of sampling collecting and take any questions.

Questions and Answers

DR. WAGES: Dr. Gray, if you were going to, how would you go about in a flock of birds, or a herd of cattle, or a farm of swine, how would you determine, go about determining their salmonella status being positive or negative?

DR. GRAY: Positive or negative in that situation would probably -- the best case scenario would be to sacrifice animals and look at deep tissue infections.

DR. WAGES: How many of those animals do you need to sacrifice?

DR. GRAY: It would depend on the species. It would depend on the flock situation, how many birds were in that barn, and what serotype of salmonella you are hoping to get a positive or negative status on.

DR. WAGES: Okay. And you maintain that there is no such things as a negative flock or herd, is that what you said?

DR. GRAY: I maintain that it is probably more difficult to find a negative flock than it is a positive one.

DR. WAGES: Because that is different from what you stated earlier that there was no such thing.

DR. GRAY: What I said was -- would be an anecdotal conversation between a student and myself if they tell me a herd is negative simply by going out and taking fecal samples. What I would say is you probably have not looked at that hard enough. Salmonella is probably there in one form or another.

DR. HOLLAND: This holds true for cattle herds too, beef, dairy? Are you just talking about pigs and poultry?

DR. GRAY: Are you talking about a pen, or a herd, or cow calf, or feedlot? I mean, those things all differ.

DR. HOLLAND: I am talking about cow calf and a reasonably well-restricted dairy herd, where you do not have animals coming and going.

DR. GRAY: What type of dairy, dry lot?

DR. HOLLAND: No, not a dry lot, free stall, close, relatively close, except for feed materials coming and going.

DR. GRAY: Those kinds of measurements are not in the literature. But if you look at what NARMS shows, herd prevalence hovers around 8 to 12 percent. So you predictably do have negative herds. But, again, you have got wildlife in the area, and that is going to effect whether or not you can find salmonella.

DR. HOLLAND: I guess I have seen too many naive beef and dairy herds to say that you have not looked hard enough to find it, both serologically and culturally, culturing for it.

DR. KOCHEVAR: Given the confounding nature of changes in assays where the percent prevalence of a particular pathogen has changed over the years in the literature because the analytical method has changed, do you think we have any basis at all to say that the load has gone up, in general, in herds -- herds, flocks, whatever? Or, at this point, do we have no way over time to have really an insight into that?

DR. GRAY: I am not sure I have the information to answer that question. But I think over time, if you look at NARMS studies, typically, the methods that have been used to measure salmonella prevalence has changed from year-to-year because those studies are not done on a yearly basis, and not done by the same lab. It is not a criticism of the study. It is a reality of those type of studies when they are done.

DR. KOCHEVAR: So, given those realities, do you think we have a basis upon which to say it has gone up or down, or we just do not have the data to say it?

DR. GRAY: In my opinion, we do not have the basis to say.

DR. WOOD: With regard to shedding, do all serotypes of salmonella not shed all of the time? I mean, is that a general characteristic of salmonella that a bird or an animal could be carrying salmonella and not shedding it?

DR. GRAY: Are you asking are there serotypes that --

DR. WOOD: Are more likely to shed than other serotypes?

DR. GRAY: It depends on the host species, and what serotype you are talking about, and those studies are hard to come by. We have to really go by field data. And an awful lot of the data we have in animals, is clinical data, because there really are not that many non-clinical infection studies done.

But one would often expect things like typhimurium to be shed more frequently than some of the other serotypes if you simply look at the surveys that have been done.

DR. WOOD: But, still, even given these characteristics, the bug of choice, if there were to be a choice, is still to be salmonella as opposed to campylobacter pathogen for measuring pathogen load?

DR. GRAY: Yes, I think that that is a true statement.

DR. WOOD: And then, in terms of confounding factors and stress, you said that there would be a need to address stress in a predictable way. I am getting feedback here. But what kinds of elements then would have to be put in place to measure stress in a predictable way?

DR. GRAY: Well, I am not a stress physiologists. But I think that Dr. Isaacson will show some interesting data on stresses that have been done, and how their outcome, what their outcome is, and that the way in which oftentimes stress has been measured in salmonella studies does not create a predictable stress, and therefore is not necessarily a reproducible way to do it. And I think if that were a factor to come into play, a stress physiologist should be consulted on that.

DR. WOOD: But stress could confound then the findings of a pathogen load studies ---?

DR. GRAY: Sure, absolutely.

DR. WOOD: Right.

DR. LANGSTON: I noted that you mentioned something that I quite often hear relative to antibiotics promoting the carrier state, and I presume that means an antibiotic to which the organism might be susceptible but has grown resistant or otherwise.

And I oftentimes quote that, but I have never actually gone back and looked at the evidence of that. Can you summarize that? Are you familiar with it?

DR. GRAY: You know, most of the studies that we use, there are some studies in swine that are quoted in the diseases of swine book that look at that. But the main ones that are used are in humans, and we know that antimicrobial treatment in humans can indeed induce a carrier state. And, oftentimes, we are assuming that that same effect is going to occur in animals with a similar serotype of salmonella.

DR. LANGSTON: And that is, in fact, when the organism is susceptible to that antibiotic, it still promotes the carrier state?

DR. GRAY: I would have to go back and look at the studies directly. But, yes, I believe that to be true.

DR. PARKHURST: Could we look at your slide number nine, the second one on measuring effects?

DR. GRAY: Which slide was it?

DR. PARKHURST: I think it was slide number nine, way back in the begin -- at -- okay, come forward a little bit.

DR. GRAY: Okay. Is this slide one?


DR. GRAY: Okay.

DR. PARKHURST: That one.

DR. GRAY: Okay.

DR. PARKHURST: Could we just start off, as a matter of curiosity, what is the Y axis?

DR. GRAY: The Y axis if CFU of salmonella per gram of feces.

DR. PARKHURST: Okay. So that is a response, a measured response?

DR. GRAY: Yes.

DR. PARKHURST: And day one is the first day of challenge?

DR. GRAY: Day zero would be day of challenge; and day one would be first day of measurement of the salmonella shedding.

DR. PARKHURST: So could you just give me a little information about at what point of intervention, what is the pros and cons at different points of intervention?

DR. GRAY: Well, I do not know that I can give you pros and cons. But the things you need to consider is if, for instance, we know that if a drug is given prior to salmonella ever being induced, what we have to -- and it has an effect on this curve, what we have to assume is that drug is having an effect on the flora in the intestine.

So it is preventing salmonella from infecting or making the intestine more prone to infection by salmonella, if it has an effect on this curve. If we give a drug late in the cycle, what we assume to happen is if we have very few organisms there, and we have a lot of other organisms that the drug can have an effect on, that we may induce carrier state by the salmonella wanting to, in a very general term, hide itself from the drug and go into the lymph nodes, go intracellular.

So, if the drug cannot reach the salmonella intracellularly, it has the ability to do that depending on the serotype. So it sort of depends on where in the cycle -- and, again, this is a clinical shedding cycle.

In a normal herd status, we probably are not going to see this shedding curve among the whole herd. We are probably going to see a low level of shedding throughout the herd over a period of time in non-clinical type state.

DR. GLENN: I have a question regarding in your conclusions you made remarks regarding study design to model a realistic infection cycle which we have been talking about of course. Then, under Study Design, you were relating these various dose response curves relative to the dose that you were challenging with.

And, as opposed to this control of natural infection, are you advocating that this control using the seeder animal is a way to get at what you call a more natural or realistic sort of scenario? Is that a type of study that --

DR. GRAY: I am implying that that is an alternative because the dose response curve that we see with direct inoculation does, in fact, create problems with -- it is not necessarily what we expect to see in the real world because our doses and our shedding do not necessarily match.

DR. GLENN: Okay. And I wondered if you might speculate on how the naive animals became infected from the seeder animal relative to your remarks on consuming feces? Do you have any idea?

DR. GRAY: I think that there are a number of ways that one needs to look at that, and it is complete speculation. I think that there is everything from the state of the organism being shed from the animal, being different than what we grow in the lab.

I think it could be that a minimal level of coprophagia in that state could occur. We also know that with salmonella intranasal or respiratory exposure can cause infection. Exposure of the head associated lymphoid tissue can cause infection. And we do not know the doses at which those things need to occur to cause an infection.

DR. GLENN: It is not related to the animal handlers?

DR. GRAY: In this case, it would not related to the animal handlers.

DR. GLENN: One last comment. How many studies of this type have you personally -- have you conducted in this whole area of assessing the salmonella reservoir? Is it predominantly with swine, I take it?

DR. GRAY: Yes, I have done a number, depending on whether they were non-studies or otherwise, sitting here counting, a couple of dozen.

DR. GLENN: Yes, okay. Thank you.

DR. KOCHEVAR: I am just curious on the shedder animal, has anybody sort of done shedder animal curve to see how low the shedder animal has to be before you do not get infection of the surrounding animals?

DR. GRAY: No, that has not been done to my knowledge.

DR. PARKHURST: It is a brief question. But once an animal is infected, does the animal become well again after a period of time, or is it always a low grade infection?

DR. GRAY: No, what you typically see in normal dose of animals is a number of animals will clear the infection, and you cannot find it again. And it is a relatively low percentage that become true carriers.

MS. SINDELAR: Mike Goodman is our next speaker. But with great thoughtfulness for everyone here, Dr. Langston has recommended a break at this time.

And so, I will passing out background material that has been prepared by Dr. Goodman to the VMAC members, and the participants are more than welcome to pick up copies that are at the back of the room. Thank you.

(Whereupon, a short recess was taken.)

DR. SUNDLOF: Our next speaker is Dr. Goodman from Exponent, who has conducted a literature view for us on pathogen load. And Dr. Goodman is here to present the results from that literature review. Dr. Goodman.

Exponent Literature Search

by Dr. Mike Goodman

DR. GOODMAN: Good afternoon. It is a pleasure to be here today. I should probably start with a previous claimer that I am a human epidemiologist, and probably do not necessarily speak the language that the guests and the members of the committee speak. So, sometimes, if it seems like it is over my head, please forgive me.


We were contacted back in mid 2000 to evaluate the published literature on pathogen load in food producing animal. And this first slide probably is more for people like myself, who tackle issue which was fairly new to us.

Our group is very experienced in reviewing literature, conducting comprehensive literature review, understanding the body of the literature in a balanced way. But this specific topic was very new and intriguing to us. Therefore, it is more for my own convenience than for members of this audience obviously.


The concern, of course, is the animals carrying increasing amounts of pathogen at the time of slaughter -- may have an increased amount of pathogens due to antibiotics or when they receive antibiotics mixed with their feed.

The task to us was to review the body of literature, published literature available to date and see what are the common themes that emerge from review of that literature.


We have started it based on an usual algorithm over scholarly review by identifying database of the literature that could be helpful for our needs. We found 33 literature databases from various areas of human knowledge, medical, agricultural, food literature, veterinarian, and, of course, general scientific. Using a variety of terms that, just given here as examples, we are able to identify a total of about 30 articles.


And then extracted information from each study using the extraction criteria. First of all, with species, we evaluated what was the antibiotic in question, what was the dose, how was the study designed, the bacterial species evaluated, and, finally, what were the findings.


It is always very useful once you collect a body of literature to try to classify it, to make the thinking process or analysis process easier. This diagram shows our current understanding of what that published literature looks like.

The total number of studies is 29; 22 of them are challenged studies or experimental studies; and only 7 of them what we termed as observational studies. In other words, there was no challenge with bacterium used.

For those of you who are mathematically inclined, let me point out that it is not a mistake; 16 + 8 + 2 does not equal 22, but the problem is -- it is not a problem. The explanation is that some studies evaluate more than one species. Therefore, the sum total is more than 22.

And then once we have understood the structure, you know, we have sort of identified the skeleton of the literature that exists, we have analyzed a study using the same box, the same criteria. And some of the important observations are that the number of antibiotics evaluated is fairly limited, less than 10.

The majority of studies that use challenge, use challenge with salmonella typhimurium as inoculum. What is also important that when you look at the years of publication, you identify a gap, and I have no explanation to it. Maybe one of the members of the committee or the audience can help us out with that.

The first studies appeared in 1953; then there was a fairly active research that appeared in peer reviewed literature all through mid-80s or so; and then for about 15 years there was nothing in the literature until '99/2000, when some additional papers started to appear.

Whether that would mean that, for some reason, no work was performed during those years, or maybe the work did not get into a peer reviewed publication, but it seemed like a drop in interest in the topic.

Using classification that I just presented in the organizational chart earlier, let us review the results of studies. The challenged studies -- and we combine them for swine and calves. It may or may not be a legitimate way to combine studies, but there are very few of them.

They really found no evidence of consistent increase in salmonella shedding in these animals with and without antibiotics. What I would like to refer you to is the summary tables that were prepared as an additional handout for this presentation.

These are too busy to be imported into Powerpoint. And, therefore, I would like you just to refer to these tables as we go along. So this slide would correspond to table 1 in the handouts.

The only study that was of interests were the 1978 publication by Williams where they used two types of salmonella challenge species; one resistant; and the other one is sensitive. And they used chlortetracycline as antibiotic of interest.

Using that dichotomy, there seems to be a discrepancy of results. The sensitive strain showed a decrease in shedding in animals that received antibiotics, and the resistance strain predictably showed an increase in shedding.

To help you understand the keys of this table, please refer to the footnote where the symbols are at the bottom. The less sign means that experimented animals. In other words, those receiving antibiotics had less shedding; the more sign, the opposite; and then when they were similar or no significant difference, there is a little wavy line.


Moving onto poultry studies, first of all, what we have to point out, the number of study was much higher. It was much more studies conducted with the poultry challenge than with large animals.

There seems to be a consistent story presented by one group of researchers that came out United Kingdom that showed significant increase in salmonella shedding in chickens that received antibiotics.

The numbers were higher, as well as duration of shedding was higher, and the results were particularly strong for avoparcin. However, it seems like these studies were somewhat in isolation compared to similar studies conducted elsewhere, say, in Scandinavia.


With regard to avoparcin and salmonella shedding studies, what Scandinavian studies found is that the increase in shedding was observed only with single inoculation in early life. However, in circumstances where series inoculations were used, and in inoculations that were performed later in life, there seemed to be now impact of avoparcin.

Similarly, another later study by Holmberg et al, found no effect of avoparcin on salmonella shedding. However, a combined use of avoparcin with monensin seemed to have resulted in an increase in salmonella shedding.


There were other studies that looked at a similar design, but for different antibiotics. These are all summarized here. And they, more or less, found no evidence of effect of these antibiotics on pathogen load.

Interestingly, these studies were also used different microbial species for challenge which makes them more interesting. In addition to salmonella, there was a challenge with E.coli campylobacter and clostridium.


The observational studies, work with term observational studies, is obviously the ones that did not involve a purposeful inoculation of animals; and then the species of interest, the ones that were looked at, were more diverse.

These are summarized in the table number three in the handout. And you can see that an interesting finding is that of effective penicillin. If you flip to table 3 of the table, first of all, most studies are fairly early. They are done in the '50s, and the latest one was done in 1960.

All three were done in pigs; all three used penicillin. And whenever penicillin was used, the shedding seemed to have increased. However, the same results were not observed with other antibiotics. I am referring specifically to two studies by Bridges, '52/'53, and a study by Fuller.

Well, this is, in a nutshell, the results of our findings. I have to say that what we also did, in addition to our own research, we identified authors of the most recent studies and contacted them to conduct a peer review, make sure we have not omitted any major studies, and their input is also incorporated in our final report that we submitted to CVM. I believe that was late year 2000.


What are the limitations of the study that we reviewed?

First of all, the only salmonella studied was studied extensively. Other findings are sort of sporadic, and there is no consistent story that emerges from reviewing those studies. As already pointed out here, earlier challenge study may not represent real life conditions.

With regards to annual species, only swine and chickens underwent a substantial number of studies. The data are lacking for other species of animal. And then, an important consideration is variability of genetic lines as diets of animals around the world.

So our understanding is that diets used in Europe may not necessary be the same as those in North American, and therefore their effect on bacterial shedding may be quite different. They create different conditions in the gut.

These are limitations of the study themselves, but it is important limitation of this review. One, of course, is that it is limited to published literature. Therefore, an assumption that published literature reflects the body of knowledge that exists out there may or may not be true depending whether or not we have publication bias at work here.

Secondly, chances are that since we have conducted this review about a year and a half ago, there have been other studies. I have not revisited this issue since about November 2000.

And, thirdly, of course, there is always the chance that we missed a study somewhere, although we certainly failed to identify any additional studies, and the type of search we conducted was pretty extensive.

Another important issue is that these studies are limited to English literature studies. They may have been foreign language studies not translated into English that just missed the search.


What do we conclude?

And I have to say that each conclusion, you may see those are the conclusions of an outsider who is used to reviewing literature, but not necessarily reviewing literature on this subject.

It is generally the theme that seems to emerge from the reviewing of these studies that the published literature at least does not seem to reflect that there is a consistent association. Now there is no consistency between studies.

The reason a disagreement in salmonella challenge studies done in poultry, an important observation that almost all of the positive studies sort of seem to come from the same group, from the same lab. That is always a bit of a red flag, because you would prefer to see independent researchers repeating each other's result, replicating results.

Overall, I would decline to draw any firm conclusions one way or another based on the published literature. What probably needs to be done is a next step is expand this review beyond peer reviewed publications and look at unpublished literature, technical documentations, and so forth.

DR. LANGSTON: Just a reminder to the committee, I have been told that some people in the back of the hall are having trouble hearing us. So, please speak up into your microphone. Any questions for Dr. Goodman?

Questions and Answers

DR. KOCHEVAR: In any of the studies that you reviewed was data presented that linked increased pathogen load to increased incidence, or even increased risk of human disease? Was that addressed in any of the publications that you looked at?

DR. GOODMAN: No, this was beyond our scope. This would be something that interests me personally much more than the scope of this project, but I think it is a different animal all together. This sounds like a great question. I do not have the answer at this point.

DR. KOCHEVAR: But with the search terms that you used, do you think you would have found that literature if it were there?

DR. GOODMAN: I have to say that 30 papers that are presented here are the ones that meet the inclusion criteria for our review. We reviewed hundreds of pages prior to narrowing down to these.

I do not recall seeing a study that I would say it seems like addressing an issue beyond the scope. I do not recall seeing a study.

DR. GLENN: Just to confirm all of these 29 studies, I believe, were sub-therapeutic use, correct? I mean, I can see that in the table.


DR. GLENN: Okay, just want to make sure.

DR. LANGSTON: Any more questions?

DR. ANDERSON: Just one question. In an attempt to try to get at the extent of publication bias, do you know who were the PIs on these studies? Did you do any sort of grouping of that? Were these academic scientists? They were from the federal government? They were for industry?

DR. GOODMAN: This is just my recollection from reviewing the papers we have. This was not one of the data extraction criteria, but my impression is 50/50. And the more recent studies, we tried contacting researchers for the peer review and those were academics.

DR. WADDELL: You stated that all of these studies were sub-therapeutic, but the doses you just listed certainly are not consistent with what we would consider sub-therapeutic. What about the duration of treatment?

DR. GOODMAN: Yes, this information can be extracted from studies. With regards to whether or not the dosages were consistent with today's understanding of sub-therapeutic, I can see there may be a disagreement. But, nevertheless, what we are looking is the intent of using sub-therapeutic doses.

With regards to duration, I will have to go and go through actual papers. It has been awhile since I looked at them. It certainly can be extracted from there.

DR. WAGES: I am sorry. I am confused now. You are saying that all of the literature that you reviewed that is in this document are sub-therapeutic uses?

DR. GOODMAN: Or, at least, that is the intent, the intent.

DR. WAGES: And how did you classify that? When you say that was the intent, who guided you on what levels were therapeutic or sub-therapeutic?

DR. GOODMAN: This was with my consulting with Dr. Matthew at the University of Tennessee, who actually is the author of one of the papers, 1999.

DR. WAGES: Because, clearly, some of these levels are therapeutic and not sub-therapeutic, not even close.

DR. GOODMAN: Well, this is something that certainly I could answer that question on a treatment in children, but not on food producing animals.

DR. GLENN: The route of administration for all studies was via the feed, correct?

DR. GOODMAN: Mixed with feed.

DR. GLENN: For greater than 14 days continuously, and they are high levels, some of them. Yes, I agree.

Isn't that our definition of sub-therapeutic, greater than 14 days? Didn't I see that somewhere today?

So, my point is that the studies appear to have the route is through the feed, albeit, some of the levels are high, but it was a continuous thing, at least probably dosed over for over two weeks continuously via the feed. So I am just asking that.

DR. GOODMAN: You know, I would probably just defer that question to somebody in the audience.

DR. WAGES: For the committee's benefit, there is in the back of our book some of the studies that they did look at. You could probably get the answer there.

DR. KOCHEVAR: Do you happen to know on the -- and I am probably going to not pronounce it.

Is it Huffton or Houghton Laboratories in the United Kingdom, the laboratory from which six studies showed significant shedding but within those studies were not able to be replicated?

I was just curious if you know whether that is a government lab, or an industry lab, or does anybody know that?

DR. GOODMAN: Yes, I need to go back.

DR. KOCHEVAR: That is okay. It is a minor point.

DR. GOODMAN: I know that the authors are the same. It is always Green and Smith in various combinations.

DR. WOOD: So it is not known though how many of these studies might be related to 558.15 then, right?

DR. GOODMAN: I am sorry. There was some noise. I could not hear you.

DR. WOOD: It is not known how many of these studies might be related to fulfilling the requirements of 558.15?


DR. WOOD: No. And would there be a different conclusion drawn -- I mean, you know, we go to the last slides, the last page of the report and look at the conclusions that basically say that overall there really is not much impact of antibiotic use on pathogen loan.

But if you look at some of these reports individually you may come to a different conclusion, particularly, if you are looking at it in terms of a particular drug to that bug, you know, as it would be within NADA, as opposed to what this report does, which summarizes findings all together as one piece. And might there be different conclusions at that point?

DR. GOODMAN: You know, my impression is that there are very few drug bug combinations, if you will, that seem to have been repeated throughout -- more than once, or say more than twice.

For those that did exist, avoparcin and salmonella, there seems to be some disagreement. So my impression is that there is no theme that really emerged from reviewing that literature.

DR. LANGSTON: I do not know why we are getting feedback on our mikes. You might just check if you cell phone or a pager or something like that that might be one explanation; otherwise, I have no idea why we are feeding back.

MS. SINDELAR: Sorry. Thank you, Dr. Goodman. With time constraints at hand for certain parties, we are going to change the schedule a little bit. And our next speaker will be Barbara Masters from USDA, and Karen Hulebak will be coming forward as well.

USDA Pathogen Reduction Measures

by Dr. Barbara Masters

DR. MASTERS: Good afternoon. I appreciate the opportunity to be here. And Karen Hulebak will be joining me later in the presentation to provide some of the information. I certainly respect the committee and the challenge that you have at had.

So, basically, we were asked to provide some information to you about the agency's Food Safety and Inspection Service's final rule related to pathogen reduction in HACCP. It is a very brief overview of how we got there, and kind of each component to that regulation, and how it has been implemented by our agency.


Basically, our pathogen reduction HACCP final rule does require that establishments develop and implement written sanitation standard operating procedures; that they develop and implement HACCP systems; that for all slaughter establishments that they conduct routine generic E.coli testing; and that we, as an agency, are conducting salmonella sampling with salmonella performance standards that we have expectations that be met for slaughter establishments and ground establishments.


Our pathogen reduction HACCP final rule was a long time in the making. After we proposed, we came back with a final regulation, and we had a fairly extended implementation process.

All establishments were required to meet the sanitation components of the reg., as well as the E.coli generic testing in '97; and then we staged implementation of the HACCP and salmonella criteria, performance standard requirements over three years.

We started with large establishments with greater than 500 plant employees in '98; went to the establishments with less than 500, but greater than 25 employees in '99; and then we implemented in all of those establishments that had less than 25 employees in 2000.

So, currently, all of our federally inspected establishments are meeting all the different components, the sanitation, generic E.coli testing, HACCP, and the salmonella performance standards are all in place in every size establishment.


What caused our agency to go with this new regulatory approach?

I think most of you were recall back in '93, the outbreak of E.coli O157:H7 on the west coast, certainly brought a lot of attention, put our agency at the forefront, and a lot of questions were raised about the effectiveness of our organoleptic system, and how effective that was in addressing the major cause of food-borne illness which, as we all know, is pathogens. So the agency went to work to address that concern and that criticism and we ended up with the pathogen reduction HACCP approach.


Having said that, in development of our final rule, it became imperative that our most important objective in putting this rule together was that we would create a rule that would allow industry to build into their food production practices, as well as, us, as an agency, to build into our oversight in our regulations effective measures to reduce and pathogenic microorganisms of raw meat and poultry products.

So that was kind of the foundation of the guiding principal in putting together our final rule for pathogen reduction HACCP, is to ensure that the industry had a process and, we, through our oversight, had a process to ensure that we were in fact impacting on pathogen reduction rather than focusing on the organoleptic inspection that we had been doing.

I think as I walk through, I am going to walk through each of the separate components to the regulation. I think you will see how they work in concert together to reach this end goal.


The HACCP requirements, for those of you not familiar with HACCP, HACCP is hazard analysis critical control points. Basically, HACCP is implemented by the industry. The industry puts together and puts in place a system of process controls, and they are intended to prevent hazards to the food supply, and a tool for the industry to use to control, reduce, and prevent pathogens in meat and poultry.

Obviously, it is not exclusive to meat and poultry. HACCP is used in a lot of the industry. But, based on our regulations, this was the intent of HACCP and the meat and poultry establishments.


The specific regulatory requirements we have in place, basically, the industry is expected to make a flowchart of their process, and look at every step within their process, and do a hazard analysis of each of those steps in their process to determine which hazards might be introduced at each step in their process.

While we, as an agency, only have regulatory authority at these slaughter and processing establishments, we do expect the establishments to consider hazards that might be brought into their establishments such as on live animals, as well as hazards that might be introduced when it leaves the establishment such as in the transportation process, et cetera.

So they are considering each of those hazards as they do their hazard analysis. They are also considering the intended use or consumers of their product. And, in that process, they are going to determine which food safety hazards are inherent in their process.


Once they do that, they then identify those critical control points are those points in their process that are critical to ensure these hazards are under control. They are required to establish critical limits. The critical limit should be that limit that is essential to be met to ensure that the food is safe.

They may have quality limits beyond their critical limits, but the critical limit in a HACCP system by regulation should be that limit that must be met to ensure the food is safe.

So, if I need to cook my food product to 160 degrees to get an effective kill of salmonella in my hot dogs, for example, and I do not reach 160 degrees, then I might have a product that has a food safety concern.

I, as a company, may choose to cook it to 168 degrees for quality purposes, but my critical limit in my HACCP program will be 160 degrees or focusing on the safety requirements of my system.

The agency's regulation requires the industry to have monitoring procedures and frequencies. These should be adequate to demonstrate that the process remains in control. They have to list their corrective actions to be followed in response to any time they do not meet the critical limit.

Our HACCP requirements and our sanitation requirements are the two places in our regulations that we require, through regulation, specific corrective actions, and these corrective actions include preventive measures.

The corrective actions in a HACCP system oftentimes go beyond what is happening at the in-plant level. I talked about the hazard analysis may take the industry back to the farm level. An example might be in a slaughter establishment that as an identified pathogen associated with fecal contamination is a hazard.

If they are dealing with continuous contamination of carcasses with feces, their corrective actions, their preventive measures, often take them back to the farm level to address feed withdrawals, and those situations on the farm that can effect animals coming into the plant, and how much fecal contamination they are getting at the plant.


As part of their HACCP system, they are required to keep records. They must have verification procedures and frequencies associated with that verification. Verification is oftentimes testing of some sort, to verify that their system is in fact effective in producing a safe product.

They have to sign and date their HACCP program. And we did put in a requirement that these HACCP programs be developed by someone that has completed a course of instruction in the application of HACCP for meat and poultry.

We did not require any specific certification, or any specific course to be completed. We did lay out the guides that we expected that course of instruction to follow to be considered a HACCP trained individual.


The next requirement that I am going to walk through briefly is our sanitation standard operating procedures. Basically, the four components to our SSOPs are that the plants implement them; that they maintain the effectiveness of their program.

Again, the specific corrective action requirements are spelled out in our regulations, and there is also recordkeeping requirements with out sanitation standard operating procedures.


Basically, we are looking at a establishments to conduct both pre-operational, as well as operational sanitation procedures. Those are the steps that they take to keep their plant clean for the production of food products. And they are also required to monitor those procedures on a daily basis.


They are expected to routinely evaluate the effectiveness of their SSOP. I talked about how you would see some of these working in concert.

An example might be listeria monocytogenes, where if they are finding listeria in ready-to-eat product, that may be an indication to them that their sanitation standard operating procedures are not effective for controlling listeria in the environment, and it might be something they are doing related to their sanitation that is causing them to have the listeria showing up in their products through their HACCP system.

If their HACCP system has, in fact, been validated, and is effective to destroy the listeria, then they are probably dealing with a concern in their environment. And that would cause them to also have to reevaluate their sanitation standard operating procedures. And, again, they should revise those if they determine them to not be effective.


For corrective actions related to SSOPs, if they find that they have direct production contamination, they have to dispose of the product, make a proper disposition of it. They need to restore the sanitary conditions. And, again, to highlight, they need to prevent the recurrence.


A third component of our pathogen reduction HACCP rule is our E.coli performance criteria. I want to specify performance criteria, because a performance criteria are not specific standards that we are enforcing.

They are criteria that were based on a national baseline data. And they are criteria that the industry is expected to test for, and test against to use to determine whether or not their sanitary dressing procedure and processes are working.

They are responsible to take the samples. They conduct the E.coli, the generic E.coli sampling. It is effective in all of our slaughter establishments. So they are expected to take them. And our inspection personnel verify that the testing is being done, and that they are taking the results of that testing to go back and look at their sanitary dressing and modify their dressing procedures as appropriate.

So, rather than enforcing the standards or the numbers, they are not standard in criteria, we look at what the industry is doing as they watch what is happening with their generic E.coli numbers.


Basically, they should assess these bacterial counts. And they are basically comparing those counts against what was done in the national baseline. And they look at these results and compare it to what is going on with sanitary dressing, and take whatever actions are necessary to lower those counts. So that is in all slaughter establishments.

So I have now walked through the HACCP sanitary E.coli. Karen is going to walk you through the fourth component which is the salmonella performance standards, and then both of us will be happy to entertain any questions that you have. Thank you.

by Dr. Karen Hulebak

DR. HULEBAK: Thank you. I would like to -- I am Karen Hulebak. I would like to thank the committee myself, along with Barb, for the opportunity to talk to you about pathogen reduction and performance standards.

My remarks are intended to give you a quick glimpse of how salmonella performance standards have helped FSIS achieve the goal of protecting the public's health by significantly reducing the prevalence of food-borne pathogens in meat and poultry products.

I am going to start with some background information on performance standards, give you an update on where we are with the standards, some reasonably current data, and talk to you about where we plan to be going in the future.

As Barb mentioned earlier, FSIS issued its groundbreaking pathogen reduction HACCP systems rule in 1996. This rule established salmonella performance standards for all meat and poultry, slaughter plants, and for plants that produced raw ground product.

FSIS chose salmonella as the target organism for pathogen reduction in performance standards for several different reasons:

First, salmonella was the most common cause of food-borne illness associated with meat and poultry products; second, it occurs at frequencies that permit changes in its occurrence to be detected and monitored; third, testing methodologies are available to easily recover salmonella from a variety of meat and poultry products; and, finally, interventions aimed at reducing salmonella are likely to have the effect, a beneficial effect, in reducing contamination by other enteric pathogens as well.


A different salmonella performance standard was established for each meat and poultry species slaughter, and for each raw product category based on nationwide prevalence studies.

For example, the standard for steers and heifers is different from the one for cows and bulls; and the standard for ground beef is different from the standard for ground chicken.

FSIS measures plant performance against the standard by collecting a series of samples to comprise a single sample set. Within each sample set, a maximum number of positive samples are allowed, and still meet the performance standard -- and the plant still meets the performance standard.

The sample set demonstrates whether plants are or are not consistently achieving an acceptable level of performance with respect to its HACCP controls.


Next, to just reiterate, the salmonella performance standards measure, are set for slaughter, and raw ground product plants. They measure process, HACCP process effectiveness in limiting contamination with this pathogen. And, finally, I should note that they are not used to determine product disposition.


As I said, the salmonella performance standards serve as a yardstick for FSIS to measure the effectiveness of industry HACCP control. They provide industry with objective, measurable standards that can be used to calibrate their HACCP systems.

And, most importantly, these performance standards help FSIS achieve our goal of protecting the public's health by significantly reducing the prevalence of food-borne pathogens in meat and poultry products.


So far, we are pretty pleased with what the salmonella performance standards have been able to achieve in concert with the other provisions of the HACCP rule. The latest complete progress report, April 2001, shows that the prevalence of salmonella in raw meat and poultry has decreased significantly since the implementation of HACCP in '98.

These are the first aggregate data I am going to show you on all sizes of meat and poultry plants. And let me quickly review them for you; broilers, average salmonella prevalence of about 10 percent under HACCP, as compared to 20 percent before HACCP; market hogs, 7 percent under HACCP, compared to 8.7 baseline; cows and bulls, 2.1 percent under HACCP, compared to 2.7 baseline.


Steers and heifers, .3 percent now, compared to 1 percent prior to HACCP; ground beef averages 3.7 under HACCP, compared to 7.5 percent pre-HACCP; ground chicken, 14.4 percent, compared to 44.6 percent baseline; and ground turkey, 29.7 percent under HACCP, compared to 49.9 percent baseline pre-HACCP.


In addition to these data showing declines in the prevalence of salmonella on various raw meat and poultry products, the Centers for Disease Control and Prevention has reported reductions in the incidence of food-borne illness.

FSIS believes the performance standards working in concert with HACCP are one of the factors contributing to these food-borne illness reductions. While all of this is good news, we know that there is more room for improvement.

This is why two activities are currently underway to review salmonella performance standards, what they have accomplished, and how they might be improved.

The National Academy of Sciences and the USDA National Advisory Committee on Microbiological Criteria for Foods are both studying the issue of microbiological performance standards.

Performance standards will remain an important part of HACCP for FSIS. Exactly what those standards should be, and how they should be applied, are questions that the agency seeks answers from the scientific community.


And, next, just to summarize and bring salmonella together with E.coli performance criteria, we see the standards, the salmonella standards, and the E.coli criteria as complementing one another.

E.coli testing is a good indicator of fecal contamination. But it is not directly correlated with salmonella contamination which is affected by a number of other factors as well including the condition of incoming animals.

The salmonella standards will continue to be used by FSIS as an indication of which plants are not currently meeting the standards, and which plants therefore need to be making -- need to be taking further steps to reduce pathogens that can cause food-borne illness.

I wanted to leave you with one final note as well. You have some very big questions in front of you, and answering them is going to take some hard work. And I wanted to note for you that the agency has, as you may or may not know, developed a risk assessment of farm to table process risk assessment for E.coli 0157:H7.

When I say farm to table, I mean to say, that the risk assessment models, the entire farm to table process of the production of ground beef from slaughter to consumption, and looks at the effect, potential effect, of various interventions on the load of pathogens in a final product, and estimates the likelihood of illness that might occur from consumption of that product which actually speaks to I think a question one of your members asked earlier this afternoon.

Indeed, the model is anchored in epidemiologic data for o157:H7. Now while the model has not been constructed exactly to answer the kinds of questions you may wish -- you might theoretically wish to ask of it, we believe that it could be used, at least theoretically, to ask more quantitative questions about impact of income pathogen load on likelihood of human disease -- income of incoming pathogen load in an individual animal, or a lot of animals on output or pathogen load in a final product.

Thank you very much. Barb and I both thank you. I will conclude my remarks, and we are happy to answer any questions you might have.

Questions and Answers

DR. KOCHEVAR: Can I assume from what you just said that there is not data available yet from that system?

DR. HULEBAK: That is right. The model has not been used ever to ask that specific question, but it is a public document. It is not complete, not that any risk assessment every is, but it is available to the public. It is out, and, in fact, currently undergoing a rigorous peer review by a committee of the National Academy of Science; ipso facto, it is a public document.

DR. KOCHEVAR: So that sounds like a very comprehensive process that is going to go on there. Short of that, is there any evidence from FSIS where USDA conducted studies that, if animals come in with a higher pathogen load, just to the slaughter plant, discounting for now conditions on the farm completely, there is no data at this point that shows that those animals create a higher risk for contaminated in-product from that plant?

I mean, that is the kind of the thing that will come out from these other studies, but it is just not there right now.

DR. HULEBAK: Right. That, again, is the kind of question we have not asked of that model.

DR. KOCHEVAR: Okay, thank you.

DR. LANGSTON: I would like to compliment what you have accomplished with HACCP. I think it is wonderful. I thought the comment on the temperature on hot dogs was very interesting, but there are still, of course, big issues.

For example, E.coli 0157 in hamburger; listeriosis in cheese, sausage; salmonella in eggs, all of which could be addressed by gamma irradiation which is an improved technique. Yet, I have yet to see any information in the press or elsewhere explaining to the public that it does not make the food radioactive.

It actually prolongs the shelf life. And if used to eliminate the pathogen rather than sterilize the food, per se, in essence, a pasteurization, it does not effect the taste, and presumably not the nutritional value.

Two questions: (1) Have you looked at, included irradiation in your risk assessment, and what that would do for those pathogens? And (2) If what I have said is correct, why has there not been a consumer education program?

DR. HULEBAK: I will take the easy one first. And that is I am not sure that we have looked at irradiation as an intervention in the process risk assessment for E.coli 0157. It certainly could be done.

Second, as the regulatory agency that approves such interventions for use on raw products, which we have done, we feel that -- and I think the rules are quite clear, that we cannot be in the business of promoting one intervention over another intervention.

We believe that it is the job of other agencies such as the Centers for Disease Control to make that kind of a public education message very clear. And I believe the CDC has said that in more than one venue that it believes that kind of message needs to be gotten over clearly to the public.

You can understand. There is sort of a conflict of interest for us as a regulatory agency approving interventions, one range of interventions. We cannot mark it one over another. But CDC, which does not have that conflict of interest in the way that we do, has said that it is interested in making those kinds of points.

DR. LANGSTON: So I realize milk is a little bit different, because it is handled at a state level. But you do not require pasteurization of milk?

DR. HULEBAK: We are not in the milk business at all.

DR. LANGSTON: So that is a state requirement, I see.

DR. HULEBAK: Well, it is also FDA's billy wig.

DR. WOOD: A question was asked earlier about the data coming in the front door, in terms of pathogen load, and going out the back door of the processing plant, I think is the question of the day.

I mean, is there any data available at all that we could look at related to that question that you were talking about that was in the risk assessment on E.coli, in terms of the pathogen load walking in the front door of the establishment on the hide of the animal?

And is there any data available on that, that is available?

DR. HULEBAK: There are voluminous data that were used as inputs into that risk assessment, and I am sure they could made available to you.

DR. WOOD: How about the number of farms where those farms have been seen are identified by establishments as a critical control point regarding the pathogen load on the hide, and where steps then were made or taken by that establishment to work with those farms with regard to HACCP or whatever? Is there data on that?

And, if so, is it a large number of farms? Are the number of arms increasing? And is there any assessment or assessment being made on what kind of impact those HACCP plans have had on the pathogen load at the front door of the processing plant?

DR. HULEBAK: That is a more fine grain question that I do not believe has quite made it into the risk assessment type work.

Barb, are you familiar with any studies that we have done, carried out, or aware of?

DR. MASTERS: We, as an agency, would not have that data. I can tell you anecdotally that we have had some, particularly, poultry establishments that have failed to meet the salmonella performance standards; that after they have failed their third set, they have to reply to us with corrective actions.

Their corrective actions included measures taken on the farm, and they did take those measures. And they did pass a fourth set of salmonella testing; and that I can think of several example where that has happened.

But we would not have the specific information, as far as to how many farms they worked with, the specific interventions, et cetera. The industry might be able to make that available to you, the National Chicken Council, National Turkey Federation. So, anecdotally, yes, but no specific data.

DR. KOCHEVAR: So their interventions might have included other things besides the on the farm stuff?

DR. MASTERS: That is correct, yes.

DR. KOCHEVAR: So it would hard to sort that.

DR. MASTERS: And they did include other things, correct.

DR. GLENN: I have a question. I would also like to compliment FSIS on these improvements that have been seen, and in their relationship to CDC's report.

Since 1998, when we first implemented HACCP, and we were looking at the salmonella reservoir, as we are trying to look at today, you indicated that broilers had been sampled, and that there is this prevalence. Tell me exactly where you sample a broiler, and a market hog, and a cow, and a bull, steers, and heifers.

DR. MASTERS: Going from memory, I certainly can get that information to you, but the broilers undergo a whole bird rinse. So the entire carcass is put in a bag and rinsed with fluid.

Turkeys, we started out with a whole bird rinse and it became an ergonomic concern, so we switched to swabbing. And the turkeys are swabbed, and my memory is not with me on the swabbing, but they are swabbed in those locations we believed to be at the highest risk for fecal contamination. There are specific areas that are swabbed.

Swine carcasses and cattle carcasses, it is the flank area, the brisket area, and the sternum area. So we can get that information to you, if that is helpful, but they are specified and considered to be those areas most at risk for fecal contamination.

DR. GLENN: Okay, great. And that is what I wanted to know because in our definition of pathogen load, we are looking at feces. And I wondered if you were out in the lot sampling feces. I wanted to make sure that was not true.

DR. MASTERS: No, actually, the animals, the carcasses that are selected have been in the chiller for 24 hours, and typically are chilled carcasses when they are selected.

DR. GLENN: Chilled, okay. And has there been any coordination as this policy is developed with FDA, CVM, as we, across the government, emphasize a farm to table approach?

Has there ever been any liaisoning to say, you know, if load is an issue then we might want to do something closer to the fecal source at the slaughter plant?

DR. HULEBAK: That is why we are here.

DR. MASTERS: As Karen says, that is why we are here. There has been a lot of communication between ourselves, FDA, APHIS, specific to the level you are talking, I do not know about that detail. But, yes, it has been an ongoing, unified effort to try to make that effect.

DR. GLENN: Okay. And do you expect further improvements of this magnitude as we go on down the road, and we are farther out from the implementation of HACCP regarding the salmonella reservoir? I am just curious.

DR. HULEBAK: It really is kind of a crystal ball. I mean, clearly, there is a great deal of interest in performance standard as a concept in making performance standards even stronger, tools more deeply and firmly rooted in science, what that is going to mean, in terms of their structure, their focus.

What pathogens we are talking about, you know, that is anyone's guess. There are two very good groups of highly qualified scientists involved, and that is more than one hand, you know.

DR. LANGSTON: Thank you very much.

DR. MASTERS: Thank you.

DR. HULEBAK: Thank you.

DR. SUNDLOF: While Aleta is switching the slides, we are back in the normal order now. And I would like to introduce Dr. Scott McEwen.

Scott is a professor of food safety and epidemiology at the Ontario Veterinary College, University of Quelth. His research efforts is on determinance of infection with antimicrobial resistant bacteria and Shiga toxin-producing E.coli in food animal populations.

Scott has been involved for quite awhile in the whole issue of antimicrobial resistance. He served on the WHO panels that have been dealing with this issue. He was at our Threshold Workshop. And he is currently a chairman, the chairman of the Canadian effort to look at the issue of antimicrobial resistance in food producing animals in Canada.

So, Scott.

Epidemiological Evidence of Pathogen Load Effects

by Dr. Scott McEwen

DR. MCEWEN: Thanks very much, Steve, for very kind words and introduction, and greetings to everybody from the Great White North. It is a pleasure to be here as always, take part in this important public health issue.


I am mindful of the time -- and I will try to be quick on areas that we have already discussed.

Briefly, I would like to go over some of the plausible mechanisms of a pathogen load effect from different types of antimicrobial use in food animals, what I think is relevant information from non-food animal observational studies, some general characteristics of the epidemiological studies that reviewed for this presentation, a brief summary of the evidence from food animal studies that I obtained from literature.

This is basically a pub med search, plus some information I had in my own files as a teacher and researcher in this area. But I am sure a keen student came up with some extra studies that I missed.


Some of this has been touched on already. I think the first possible effect of antibiotic treatment, whether it is sub-therapeutic or otherwise, on animals could be an increased susceptibility of animals to infection, essentially, by reducing the infectious dose.

Two possible ways this could work: If there is treatment before exposure with the pathogen, that is, a previously uninfected animal, it could increase the susceptibility of the infection by suppressing normal flora, thereby diminishing the animal's colonization resistance.

The second mechanism treatment, during exposure to a resistant pathogen facilitates infection by the selective effect of resistance.


A second possible mechanism of treatment is increased duration of shedding or concentration in feces. We have talked about that, possibly due to related mechanisms to greater degree of colonization, perhaps, or attachment sites either intra- or extra-intestinal infection in lymph nodes or elsewhere, and possibly with disruption or normal flora.

And a third major effect antibiotic treatment could have I think on pathogen load is to diminish it by decreasing the prevalence or duration of shedding due to treatment.


Just, in general, we have talked a lot about experimental studies today, that is where you randomly allocate treatments to animals in groups typically. I am referring to epidemiological studies. And in today's context those are observational in nature, both from respect to exposure to the pathogens of interest, and to exposure to drug treatment.

So these both occur as they would on the farm. That is one of the major benefits of conducting such studies. The other major benefit is that multiple causal factors or risk factors can be investigated, and there is many of these that Jeff touched on very well in his presentation relating to the agent host and the environment.

And one other thing I would mention is that another important aspect of animal husbandry in today's world is it is very hierarchical in nature. And epidemiological studies that are designed properly can take into effect things like group effects, herd effects, hatchery effects, and that sort of thing.

We always have to take great care in conducting these studies to try and design them appropriately and analyze them carefully to avoid the many biases which these studies are subject to.

Some of these have been mentioned; things like, selection bias, misclassification bias, confounding, and other things. Epidemiologists spend a great deal of time trying to prevent or to explain these sorts of biases.


And I should mention that not all of these studies are the same type, and different designs have various strengths and limitations with respect to causal inferences.


A few points about some related studies in other species that I think are relevant. There has been several case control studies in humans, in particular, showing prior antimicrobial treatment as a risk factor for clinical salmonellosis, and recently a paper showing its effect for campylobacteriosis.

We have had similar observational studies conducted in horses, hospitalized horses and dogs with nosocomial outbreaks of salmonella associated with prior treatment with antimicrobials.

There is some evidence for causal role of drug treatment enterocolitis of horses and rabbits. I should emphasize that these observations involve clinical disease and not subclinical infection.


The first example is from the human realm. It involved a case control study that was described about 10 or 12 years ago, an outbreak due to an antimicrobial sensitive strain of salmonella havana.

In this study group, prior antimicrobial therapy was a risk factor for clinical disease. And the odds ratio is 4.3 indicating a fairly strong association with prior antibiotic use.


From horses, a study conducted in California, I believe, case control study of salmonella St. Paul infection in hospitalized horses. Horses receiving parenteral antimicrobials prior to the first culture for salmonella were at 10.9 times greater risk of infection than horses not being treated.


Before going on to some of the food animal studies, I think it is important to talk about both the effect measures and outcome measures that were used in these studies. It is quite a range.

Most of them measured farm level culture status, that is, the prevalence of infection at the group level, a qualitative, yes or no. Some assessed individual animal status, or the proportion of the herd that were shedding, so a more quantitative outcome.

Duration of shedding/infection, concentration of pathogen feces were not measured explicitly, although, to some extent, these weigh into a prevalence measurement because prevalence is a function of incidence, the number of new cases that occur over time and the duration of infection in the population.

One important point I think is that most outcomes were relative unrefined. Multiple serotypes were -- usually no attempt to differentiate serotypes or other important agent or host characteristics.


Similarly, most studies had somewhat aggregated exposures, measures, or risk factors. Most measured treatment at the herd level, which is often appropriate if we are talking about sort of group treatment effects, were difficult for individual treatment.

Growth promotion, yes or no, in some cases, specific drugs were mentioned, others not. And, sometimes, they quantified therapeutic. Was therapeutic drugs administered to this flock or herd? Yes or no.

If you looked at individual animal treatment effects, usually not drug-specific, none measured the duration of treatment, and most treatment variables were unrefined, as I have mentioned.


This slide is a summary of the results of these studies presented in a qualitative nature, a format. The number in brackets represents a number of studies over the last 20 years or so, which looked at pathogen load effects in that species of animal, and species of bacterium.

The NS means it was not significant The drug did not significantly associate with pathogen shedding; minus sign indicated a protective effect, or a negative association; and positive, the opposite.

In poultry, three studies looked at salmonella epidemiologically; two, there was no significant difference between treated and untreated animals. In one study, there was a protective effect. In swine, one study showed a positive effect of treatment; two showed no significant difference. In cattle, similarly, one showed a protective effect; two, no significant difference.

With respect to E.coli 0157:H7 in cattle, four studies were in the literature. One study showed a protective effect; one a sort of positive effect, which I will talk about in a few minutes; and two, not significant. In campylobacter, two studies in poultry, one protective, one not significant; and in cattle, one not significant.


I will go through a few studies done in animals to illustrate some of the principles and findings. The first was a study of E.coli 0157 in dairy herds in the Pacific Northwest, United States. In this study, they undertook regular monthly fecal culture of cattle and classified farms as to positive or negative.

The investigators reported a tentative association of E.coli 0157 prevalence, and the feeding of ionophores in heifer rations. This was an unconditional association, and the authors did not undertake any multivariate studies on this particular outcome.


Second example, a large study conducted here in the United States. Again, cross-sectional study of U.S. feedlots. I believe it was a cattle on feed evaluation study conducted by USDA. And they were undertaking to identify risk factors for E.coli 0157 shedding.

Sixty-three of the hundred feedlots examined had at least one positive sample, and these investigators found no significant association between positive fecal culture samples and ionophore use, or with feeding of antimicrobials in these feedlots.


The third example is, in this case, a pig study. This was undertaken in the Netherlands, and involved 353 pig farms. And the outcome in this case was serological or seropositive status of blood samples of herds taken at slaughter.

In this instance, the use tylosin as an antimicrobial growth promoter in finishing feed was significantly associated with a higher salmonella sero prevalence on farms.


Fourth study, very large, well-conducted, comprehensive study undertaken in Denmark was a cross-sectional study of about 4,000 broiler flocks, 12.6 percent of which were positive on fecal culture to salmonella typhimurium.

In this study, the use of unspecified antimicrobials was associated with reduced risk of salmonella infection, specifically in flocks that came from salmonella negative parent flocks; association did not hold in flocks coming from positive parent flocks.

In this study, growth promoters was not significantly associated with salmonella infection.


In summary, I have only shown you a few of the studies involved that is really representative I think. We have only had a modest number of epidemiological studies which have looked at this issue of pathogen load, approximately 15, by my count; none assessed carcass contamination or other public health effects.

Most evaluated salmonella. Fewer looked at Shiga toxin-producing E.coli and campylobacter. And most studies, I think, importantly, sought to evaluate a broad range of potential risk factors, in essence, screening studies. None were specifically designed to assess pathogen load.


I think one sort of caveat that took place on this is, given the exploratory nature of these studies, and the comparatively unrefined nature of both the exposure and outcome variables, important associations may have gone undetected. And I think any future epidemiological studies in this area should be specifically designed to address this topic.

Another point, it may seem obvious, is that such studies are inherently post-approval because you cannot undertake observational studies until the drugs are on the market.


So, in conclusion, most studies found no evidence of a pathogen load effect; some found evidence of a protective effect; and some found a positive effect of drug use on pathogen load.

Overall, I believe that current epidemiological evidence suggests that undesirable pathogen load effects of antimicrobials used in Europe and North America, if they indeed exist, are probably minor. Thanks very much.

DR. LANGSTON: Questions?

Questions and Answers

DR. LANGSTON: I think on your slides, 7 and 8, where you give examples, one a human study, and another in horses, where antimicrobial treatment increased the odds ratio in developing salmonellosis, do you have any idea whether those salmonella were susceptible to the antibiotics that were used in either the humans or horses?

DR. MCEWEN: Well, in the human study, this strain was reported to be antimicrobial-sensitive. So, I gather that means universally-sensitive to antimicrobials tested. I do not know what range of drugs they tested. And I cannot recall from the horse study whether or not they tested the resistance or the susceptibility status of that strain.

DR. KOCHEVAR: As a follow up to that, in any of the other studies was there any consideration given of incidents of drug resistant strains in any of the questions that were asked epidemiologically?

DR. MCEWEN: Yes, I mean, I do not have any general conclusions. I explicitly left out studies that sought to evaluate effects on resistance, and tried to select those which either ignored resistance or did not measure it.

I mean, it is difficult I think to separate completely the issue of resistance from pathogen load because the mechanisms do overlap. So it is a bit artificial to do that. But, in the case of this particular exercise, that is what I did.

DR. WAGES: Scott, does Canada have anything in place in their drug approval process that requires pathogen load studies in the approval process?

DR. MCEWEN: I do not believe so, no.

DR. WAGES: If you were asked to comment would you recommend that they do so, based on the information?

DR. MCEWEN: I was afraid I would get that question.

DR. WAGES: Based on the information that you have given to this committee?

DR. MCEWEN: Well, I very well may be asked to comment. I think, based on what I know, what I have seen and heard at previous meetings, and today, and read in the literature, I think preapproval studies are problematic from a design point of view.

My preference would be to make sure that we have a fairly capable post-approval surveillance program looking at both resistance, pathogen load, and antibiotic use. That is a big undertaking.

DR. WAGES: Thank you.

DR. LANGSTON: Any other questions?

(No response)

DR. LANGSTON: Thank you.

DR. SUNDLOF: Our next speaker is Dr. Richard Isaacson, who is the with the University of Minnesota, but formally from the font of all knowledge, University of Illinois. So, Dr. Isaacson.

Factors Affecting Pathogen Shedding

by Dr. Richard Isaacson

DR. ISAACSON: For those of you on the panel, the slides that I have here are handed out to you separately. So you do not have them in the notebook, as far as I know. And that is because we are using different computer systems. And, apparently, the FDA's version of Powerpoint could not translate my MacIntosh version.


I have asked to talk about some of the factors affecting pathogen shedding, particularly, with salmonella. I was also told that there has been no levity in this meeting yet, so I thought I would start off with a little bit that might state probably some of the situation that we are going through right now which is stated here.

We are all visionaries. I am a visionary. You are a visionary. But what do you do? And so, we all have different occupations, and so we look at these things in a slightly different way.

I am a microbiologist, and we work primarily with swine systems. And everything I tell you about pretty much will be relative to swine. And I presume it is translatable to other systems, but I do not have data to support that.


So I want to talk about a number of things. And, really, this is the outline, and these are the points I want you to leave with. One is that, with regard to swine, most farms have salmonella. And one question is, but is it the right salmonella as a food-borne pathogen question?

Shedding of salmonella is sporadic. In healthy pigs, which is what we are mainly concerned about, either cecal or fecal concentrations of salmonella, post-challenge, that is after things have kind of equilibrated, is usually very low.

We are talking about between about 10 and 100 cells per gram of feces. And one question that goes through my mind is that enough to cause disease? Is that enough to impact resistance transmission? What is the relevance of that?

That, apparently, healthy pigs can be carriers, and that there are a number of factors that can be used to express the shedding of that organism from the animal; and also want you to come away with the fact that detection is difficult.

We have seen a number of presentations already this afternoon on detection, and even on some quantitative measurements. And let me tell you that those are not the simplest things to do. You cannot just go and plate it out. So I will discuss that.

Growth promoters, and I will give one study that we have been involved with, do not necessarily affect salmonella shedding.

And, finally, is that salmonella can also be rapidly acquired. So you can do all of the nice beautiful studies, and it can all be for not because rapid transmission at the every end of the study.


So let me just start off by telling you, this is some data that we got from a study that we have been doing with pigs in Illinois. And it is just to show you the serotypes that we have identified from healthy animals, healthy, mature animals in Illinois from 11 different farms.

And the point is, is that these are the serotypes that we have been seeing. Jeff Gray showed you some of this data as well. But if one looks at this from the standpoint in this particular study, you will notice that there is no typhimurium there, and that was a bit of a surprise to me.


But if we look at what is important in humans -- and you will not be able to read this, I am sure. But this is a 10-year summary of data from CDC samples, and this is typhimurium; this is enteritidis; it is Heidelberg. It gets all the way down here to, I believe, agona becomes the first on my list.

So the question is: When we go out just in the field and look for salmonella, are we looking for the right ones all of the time, or do we pick out the right ones?

I have no doubt that salmonella typhimurium is common from pigs, and that pigs are a common source of contaminant, or food contamination. So it is part of the time is that we are going to be looking at things that are not necessarily related to the most prevalent or food-borne pathogen.


What about the actual prevalence of salmonella on farms?

So we did a study looking at initially 141 farms using slaughter plant samples. And, in fact, we looked at lymph nodes. And when we looked for salmonella, any kind of salmonella in these samples, what we found initially, out of those 141 farms that were surveyed, only 65 percent of them showed at least one positive salmonella, and we looked at 30 pigs per farm.

And we even could divide those up into what we called high prevalence and low prevalence farms, thinking that this might be a way in which we could start looking for risk factors that might be involved in their being there.

In a follow up the following year, we looked at a subset of this 141, almost 100 of them. And what happened is that now many of those that were negative were now positive; some of those that were positive are now negative; some of those that were high prevalence are now low prevalence, and all of the combinations that you could think of.

And, at this point now, if we took a two year running total, 90 percent of the farms that we looked at were positive. And then, in a subsequent third year of sampling, it starts approaching 95 percent of the farms that we have looked at are positive.

And I think Jeff Gray said, you are just not looking hard enough. These are single sampling times, so we are not going back and repeatedly doing this on a specific farm sample other than once a year. So the answer is, yes, if you look long enough, and you look hard enough, you will probably find salmonella.

And so, the idea is that -- one is that most farms have salmonella, swine farms, and that shedding is sporadic. One day you can find it, and the next day you cannot.


Okay. So question about concentration of salmonella in feces. I have been quoted at times to say I would rather eat a piece of meat contaminated with salmonella, say, 102 or 101, than 107. And not that I do not know how to prepare the food to kill the 107, but I do not want to be eating that amount of fecal material.

So, in this study, what we did is we challenged pigs orally with the salmonella typhimurium. It is a wild-type strain that we know we will persist in the colonized pigs, and we followed it over time. And what you see here is the concentration and log ten, and you cannot just take and plate this out, as I mentioned.

We have got to go through a very laborious double enrichment procedure, as well as doing dilution, so that we can get a most probable number calculated, which is not the same as taking a sample, doing a dilution, and plating it on a plate, and then coming back the next day and counting.

And what we find is that post-challenge is that this is really about a week post-challenge, 56 pigs challenged, is that the level is up in the about 103 range, so a couple of thousand cells per gram of feces that are sampled.

It goes down, and then it popped up here, because we did a second challenge. We were a little concerned that we did not actually get all of our pigs, so we did it again. And you can see after that it goes back down.

In 120 days post-challenge, the actual average concentration is extraordinarily low. We are talking about less than .1 percent of the -- or .1 colony forming units per gram of feces. That is the sum of all 56 pigs.

And what that represented is about three or four pigs that were still shedding, and the rest of them that did not appear to be shedding at all. And those that were shedding were in that 101 to 102 range, and that was it.

So the concentration is very low. And I would question whether or not that is an infectious dose for a human disease without an adulteration or a breakdown in the processing procedure post-slaughter.


Okay. Apparently, healthy pigs are in fact carriers of salmonella. A similar type of experiment, what we did is we took, in this case, 46 post-weanling pigs. We challenged them orally with 108 salmonella typhimurium.

It is a strength 798 that was, in fact, isolated from a pig, and was obtained from NVSL some years ago, and we had marked it with nalidixic acid resistance just so we could identify that we had the right strain when we were done, orally challenged them, and then at four weeks interval -- actually, four weeks later we rechallenged them because we wanted to now go with the same protocol that we had established and did not want to start changing it.

All of these animals are then reared sort of conventionally, that is they were not isolated -- they were isolated from the outside, but they were grouped together in pens, so that they had access to each other during that period of time.

And they were reared until they hit about market weight, about 240 pounds; and then they were divided into four different groups. Those four groups were listed here:

Group one simply remained on feed. They were segregated off. They remained on feed. Group two had feed withdrawal. In fact, two, three, and four had feed withdrawal of 6, 12, and 24 hours; and then each group was separately transported, put on a trailer, and trucked around for about three hours, brought back to the lab, and we necropsied them and asked what was there.

Because we did feed withdrawal, we looked at cecal samples in this case. And just to show you what happens over time is that, again, post-challenge, one month post-challenge, you can see about 80 percent of our pigs are shedding, and it goes down to the five month period almost none.

I believe this is only three pigs out of 60 that were actually shedding at that particular time. This is percent positive now, instead of the actual concentration.


But when we did looked at the pigs in these different groups what happens is that the least -- these are the feed withdrawal times. The zero feed withdrawal pigs had a somewhat increased elevation in the number of pigs that was now positive.

As the time of feed withdrawal increased, the number of pigs that were now shedding, or that we could detect it in cecal content, went up. And this represents actually about 83 percent of all of the pigs.

So we can have apparently healthy pigs that are not even initially shedding -- appear to be shedding salmonella, that ultimately can be shown to still be contaminated. And, in fact, if they are stressed appropriately, they can be brought back to the shedding status.


Detection of salmonella is difficult. In fact, the question is, is culturing is what is the sensitivity of the assay? We do not have an idea of what the sensitivity of even culturing is for sure from in vivo samples. From a lab grown sample we can tell you, but not from in vivo samples.

What are the conditions? What culture conditions affect detection? Is it reproducible? What is the impact of repeated sampling?


And there was a very nice paper that was published last -- two years ago. It was a collaborative effort by Peter Davies at North Carolina State and Paula Fedorka-Cray at ARS looking at the two labs' different methods of culturing.

And at North Carolina State, Peter Davies used this method, which was to take ten grams of feces, put it in buffered peptone water, do an overnight pre-enrichment, then put it into Rappaport's or RV broth, as a second enrichment, followed by plating on indicator plates, XLT4, and then analyzing what was there. And that was their standard method.

And so, they decided they would compare it to what Paula was using, which was to use instead one gram using either one of two enrichment procedures, either using GN Hajna as a first enrichment, or tetrathyanade broth, followed either then by again the RV, and plating on XLT4.

Well, the first comparison was to simply look at the effect of sample size. And I am talking about numbers of grams of feces to see what happens. And this was just using method one, so that is the Davis procedures.


And, as you can see, one gram of feces gives you about 6 percent of all positive animals that were looked at were positive. But if you increase ten-fold, it now was 20 percent of the samples were positive.

So in trying to figure out how much, or whether salmonella is there, it really depends upon what your sample is, and how large that sample size is. But there are other confounding issues here such as: What method do you use? And which lab does it?


So here is a, in fact, a matrix of the different laboratories; laboratory one, which was the Cray lab I believe. Laboratory two is the Davies lab. Method one is the Davies method; two and three are the Cray method; and this is the summary of them all.

And you can see that if you just compare, say, method one, is that the lab that developed it got more positives with the same set of samples. Now, I should tell you they used the same set of samples, as did laboratory one -- or did -- excuse me -- yes, as laboratory one did.

Laboratory one's method, you see, they did better than did laboratory two. And, in fact, the point is is that the methods that you use can generate different results. And, in fact, some of the positives that the Cray lab found as positive were different than the positives that the Davies lab found.

So methods are important, as well as who is actually performing them. So there are variations that can occur. And trying to standardize an assay then becomes, as you can imagine, quite a headache.


Okay. We did then a study, and I want to talk then about the growth promoter aspect of it. We did not do this to find out if -- what we were really trying to find out was whether flavomycin or bambermycin had an impact in reducing salmonella shedding. And I can tell you that it did not.

But we took 60 pigs, challenged them again orally in our model with 108 nalidixic acid resistance, salmonella typhymurium strength 798, divided them after that, a couple of weeks after into two different groups. One got flavomycin, one did not; raised them again conventionally.

But now the two groups were separated from each other, did our feed withdrawal, 24 hour feed withdrawal followed by three hours of transportation because we knew that would be one way of getting an increased shedding rate.


And what we found was this kind of data. The red is the bambermycin, or flavomycin. The blue in the back is the non-treated group. This is the shedding over time. And you can see that initially, actually, this treated group was slightly lower.

But I can tell you that there are no statistical differences here. Even though there is some apparent differences visually, they were not statistically different.

And, at the end, you can see that indeed, if anything, the treated group was slightly higher; but, again, statistically not different, so that there was not really any impact of flavomycin on the shedding of salmonella in this study. However, we did show that flavomycin actually was doing something because we did have parameters of performance.


We looked at weight gain and feed conversion. You can see the treated groups had a slightly higher, 1.7 grams daily -- pounds, excuse me, of weight gain average per day compared to the non-treated group was slightly lower. And the conversion rates, feed conversion rates were slightly in the benefit of animals that receive flavomycin.

So, in fact, it was working as a growth promoter, but it had absolutely no impact on the salmonella shedding.


Okay. Then the final point that I was going to make is that salmonella can be rapidly acquired. And Jeff Gray, again, mentioned some of this earlier, and I am going to take it just a little bit deeper, is Paula Fedorka-Cray did do some initial studies, really elegant studies in '95, showing that pigs could be challenged, rather than orally, intranasally instead, and that within hours after intranasal installation that you could find salmonella in the gut of these pigs, and not just necessarily in the tissue, but actually in the lumen, so that there is a very rapid transition from respiratory tract into the gut, and they can be secreted.


Taking that a little bit further, Scott Hurd at NADC has taken, and using pigs that were challenged intranasally using the feces from those, and then putting them in the pen with naive pigs showed that if there was about 103 colony forming units of salmonella typhimurium in a gram of feces from these initially challenged pigs, and naive pigs were exposed to that, within two hours you could start seeing that pigs are becoming positive.

Now, these are the samples that were looked at. Some of them are lymph nodes; some of them are gut samples. And it just took one positive sample to be called positive in this.

The point though is that by six hours after exposure to these feces, all six animals that were exposed in that case were now positive for salmonella. And the point was that when animals are transported and brought into a start or a holding pen at the slaughter plant, lairage, that those animals can very rapidly pick up salmonella even if they were not carrying out.

So, even if the herd that it came from did not have salmonella, which is highly unlikely, they could still pick it up, or they could pick up a different serotype. And it can happen within hours.


So, the points were, most farms have salmonella, check; shedding is sporadic, yes; very low concentrations, yes; and even those low concentrations are enough to get it into a new pig.

Healthy animals can be carriers. And, in fact, apparently, non-shedding animals can oftentimes be positive. Detection is difficult and the culturing systems are very complex. At least one growth promoter had absolutely no impact on salmonella shedding, and it can be in fact transmitted and acquired very rapidly.


The studies that we did were in conjunction with a number of individuals. These are the people, and these were the sources of funding. And I will take any questions you have at this point.

Questions and Answers

DR. KOCHEVAR: I was interested, when you talked about the effect of withholding feed on increasing the number of animals that were shedding; and then I remembered what they talked about with the FSIS presentation that -- maybe I misheard this -- but that they encourage withholding of feed in order to limit the production of feces to limit contamination. So which one do you go with?

DR. ISAACSON: Well, it is a bit of a catch 22 because if you withhold feed, then the gut is empty. And that is why we did cecal contents, because we could not reproducibly get feces. And so, the potential for nicking a gut and having it contaminate the surface becomes reduced.

On the other hand, the potential for salmonella being there is slightly higher. If we looked at it from another pathogen standpoint, it may be absolutely the exact opposite.

In fact, some of the studies that have been done with E.coli, and doing changes in the feeds that they are getting, take them off without the high growth feeds and putting them onto grasses just prior to shipping, which we think would actually reduce salmonella.

In fact, I think that the studies actually showed that that would reduce salmonella -- actually, no, increase salmonella, but reduces E.coli, so but that is in cattle. So I think that the answer is going to be it depends upon what you are looking for that particular day.

It may have a positive effect for one thing, and a negative effect in the other. And so, the net result is what ends up on the carcass. I would just sort of add though from a food safety standpoint and a question of the impact of resistance transmission to humans, is whether a 102 contamination still is a relevant number, or whether we need to be thinking about something higher than that.

I think one of the problems that I have seen in the literature, and one that I have worried about a lot, is are we worrying about the wrong phenomenon?

Are we worrying about these low level contaminations as being the source of food borne transmission, or is it the occasional, very high animal that actually is the one that contributes to the real problem?

And no one is looking at that, and even with the FSIS rules, it is a yes/no. It is not a quantitation. So I think it is another question that certainly needs to be thrown into the mix and addressed.

DR. GLENN: A question regarding the same study with the time of feed withdrawal increasing -- as time of feed withdrawal increased, study increased. These animals were challenged two times, I believe, with the same dose probably both times, ten to eight

Did you have a control group? What would be your speculation on just a regular ole' healthy pig?

DR. ISAACSON: Do you mean would there be any salmonella there?

DR. GLENN: Would there be this interrelationship between time of feed withdrawal and shedding?

DR. ISAACSON: Well, we --

DR. GLENN: What do you think?

DR. ISAACSON: When we selected our animals from a university herd that we did not have -- we could not demonstrate the presence of salmonella there. And we never did find any salmonellas in these -- at any other serotype, any other non-antibiotic now resistant strain there.

So, if we had done that, we did not do that, if we had done that, we expect that we might not have seen anything. But it is an interesting question because, you know, maybe it was there hiding and we did not find it.

DR. GLENN: Yes, my concern is if the animals are sick upon challenging, then I think there is a lot confounding effects occurring relative to gut motility and the shedding rate. And so, you get this confounding effect of, you know, if you are sick, you know -- I realize you went through the feed, so that is good. That was controlled. But, still, I was just wondering about that.

DR. ISAACSON: And so, we did not look at a non-challenge group.

DR. GLENN: Okay.

DR. ISAACSON: But you mentioned sickness, and I think I should mention that if the challenged dose that we took caused about 50 percent or so of animals to have a transient mild diarrhea. None of them were ever dehydrated. The other 50 percent never showed any clinical signs.

And, interestingly, on the second challenge, as you might expect, there might be some protective immunity; the second challenge, we saw no illness at all associated with it.

DR. WADDELL: I was wondering, in light of Scott Hurd's study on the lairage, did you guys clean and disinfect the vehicle that you transported those pigs with, and hold them in a different area before they were euthanized?

DR. ISAACSON: We did not hold them. So the vehicle was actually large enough that we could put two groups on there. And there was a good solid partition, so there was no cross-contamination. We did two groups one day; and then the next day, we did the next, and it was completely disinfected during that timeframe.

When they came back, we tried to do our necropsies very quickly. So, within about an hour, we were able to do 15 of them. An interesting question is we asked whether there was anything, any difference between the ones at the very first to be slaughtered versus the 15th animal in that group, and whether there might be a pattern there, and there was none.

So we did it quick enough I think that we did not see probably the lairage issue that Scott has mentioned. But it is possible that we see some of that because it is a three hour transport, that there could be some of that going on in the truck.

But, remember, these animals, particularly the ones that were 24 hours, which were the highest, there was nothing to defecate, so there was -- if that was going on, it was -- would be a pretty interesting -- well, it would be hard to rationalize how that might be occurring.

DR. WADDELL: Were you surprised on your survey that you did not find a cholerasuis in pigs?

DR. ISAACSON: No, because when you find cholerasuis, you generally find sick pigs, and these were healthy pigs. And, generally speaking, out in the field it is hard to find cholerasuis. And that is something that kind of comes and goes. I do not think we have seen too much cholerasuis in disease settings recently. It has declined quite a bit.

DR. WADDELL: Has the avery live vaccine had an effect on long-term shedding?

DR. ISAACSON: I think there are some claims to that, but we have not looked at that. So I could not really comment on it. I think there has been some claims that it has some impact. I am not sure how much.

DR. PARKHURST: In that rapid infection study by Hurd, it appears that after six hours there were only six pigs left. Is that correct?


DR. PARKHURST: What happened to them?

DR. ISAACSON: I have to tell you. I pulled this out of the literature, so I mean it is his study. I think they only had six animals at that point in their group. Actually, I am thinking -- no, because they were actually -- they had three different separate -- you know, three separate groups in there.

And I guess I am just assuming their group size was not sufficiently large. They have replicated this several times, and they still see the same kind of effect.

DR. LANGSTON: This is more just kind of a point of interest question. I have been reading a little bit about quantitative PCR techniques. And, given the difficulties, would that hold any potential for this sort of thing?

DR. ISAACSON: Well, I think that is a good question because you could start asking the question about whether or not it is there. It is very difficult to do quantitative PCR from fecal samples. There is just all sorts of inhibitory substances there that makes it very difficult to do.

However, we have done some work, and others have done some work of doing either pre-enrichments, or doing enrichments, and then working from that to at least move the timeframe forward a bit.

I mean, to do what we were doing, the standard procedure, it takes almost a week or more, a week to ten days to get confirmation of what you are working with. That is a long, long process. For research it works okay. If it was to be invoked in the field for any purpose that would be inadequately.

So we have looked at PCR, and we have actually had some pretty good, albeit, preliminary data by rather than using the tetrathyanade enrichment, we can go right into RV, do an overnight incubation, pull a sample out of that, and then do the PCR, and it works reasonably well.

I know that Randy Singer, who is at the University of Illinois, is doing similar kinds of work with cattle samples, and he has been able to go into tetrathyanade, initially takes a sample out after overnight growth, takes a sample out, dilutes it into BHI, lets it grow for about three hours, and then does the PCR. And, again, he gets pretty decent results with that as well.

DR. WADDELL: Are programs that would be looking at elimination of salmonella from swine farms, is that a practical goal?

DR. ISAACSON: Do you want my opinion? No, I do not think it is everywhere. We have looked at pigs. We have looked at mice. We have looked at insects. We have looked at birds. We have looked at people bringing it in.

Unless you could completely isolate the animals from any contact on themselves, as well as any type of wildlife, and that includes the insects, it is going to be really hard to get rid of it.

I think it is really ubiquitous. And for a long time as a microbiologist, I would say that salmonella was a pathogen because it is inherently invasive. It likes to find M-cells and go into M-cells. It goes into enterocytes. It likes doing that.

It likes to get into even leukocytes. And I thought, well, that is pathologic. It has got to be pathogen. But the more I looked at it from a farm standpoint with these healthy animals, believing that it is an indigenous part of the microflora.

And I do not think that it would be any easier to get rid of salmonella than to say that we want to eliminate all E.coli as well. I think it is asking for something that would be really, really difficult.

Could you reduce it? Yes. But I do not think it is elimination as a practical approach.

DR. LANGSTON: Any other questions?

(No response)

DR. LANGSTON: Thank you, doctor.

DR. SUNDLOF: Okay. Our next speaker is Dr. Mark Robinson, who is the director of the Division of Human Food Safety in CVM. And Mark spoke yesterday, but for those of you who are here for the first time today, here is Mark.

Human Food Safety Evaluation of Animal Drugs

by Dr. Mark Robinson

DR. ROBINSON: Thank you, Steve, and good afternoon, good late afternoon.

As Dr. Sundlof mentioned, for those of you in the audience who were not here yesterday, the committee has already heard my schpill with regards to human food safety evaluation, as it relates to chemical drug residues. And I won't suffer that upon you today.

Hopefully, my talk will be more on the order of food for thought. And before I get to it, I would like to make a couple of comments with respect to issues that have come up, which the committee may have questions.

The first is that the 21 CFR 558.15 is still extent. It is in the regs. But in the last period of about three to four years, we have really been operating in the evaluation in the preapproval mode, with respect to the framework document and guidance 78, which, in particular, guidance 78 causes us to look at the effect of the us all classes of antimicrobials, so the therapeutics come in partly as a result of that.

The second comment is that there has been reference to a workshop that was held in February of 2000. And I would like to give my observation of that workshop. This was what has been referred to as the preapproval workshop.

The central focus of that workshop, from my perspective, and I was new. I was on-board for one week, having returned from five years in Europe. So I think I had the newest eyes in the FDA group looking at this workshop.

And the central focus, it appeared to me, was to look at the types of studies that would be revealing with respect to the rate and extent of resistance generation. And because the framework document includes both antimicrobial resistance and pathogen load as subjects of concern or subjects of evaluation, pathogen load kind of came along for the ride.

I did not hear one unanimous voice with respect to pathogen load studies. I heard a number of voices. One voice that I heard was the repeated statement that 558.15 studies and pathogen load studies in general really could not provide much, if any, information with respect to the question of the rate and extent of resistance generation, which I think is logical.

Another voice that I heard was that pathogen load studies, in the generic sense, just were too difficult to perform. There were too many confounders, too many variables, therefore, we should not be in that business.

And, as Dr. Shryock has mentioned earlier, he pointed out that the types of studies that have been done to date trying to draw any inference with respect to those studies to on farm practices, to contamination at the slaughter house, or to public health effects, is very difficult.

The voice that I did not hear at that meeting, because there was a an FDA public workshop, and the FDA was there to listen not to drive the direction of the workshop, the voice that I did not hear was the regulatory voice. And so, that is what I will try to speak to now in this presentation.


I am going to drag you through just a little bit of yesterday, which is to remind you that in the preapproval human food safety evaluation, we look at the sponsor-generated data demonstrating the effects of defined concentrations of an active ingredient, formulation component, metabolite, or drug product in relation to specific endpoints of relevance to public health, in contrast to the evaluation of chemical drug residues.

If we are talking about pathogen load, changes in bacterial drug susceptibility, or even competitive exclusion products, we are talking really about the drug product, and the potential effects of the whole gamish, because we are involving biological phenomenon not just chemical phenomenon.

We do this, again, because, in part, because of Section 512(d)(2) of the Act which states, "In determining whether such drug is safe for use under the conditions prescribed the secretary shall consider, among other relevant factors, the probable consumption of the drug and of any substance formed in or on food because of the use of such drug."

As I mentioned yesterday, that is a pretty wide open field with respect to interpretation.


The point again is to identify any potential adverse human health affect that may be caused by the consumption of the animal drug residue and edible tissues from food animals.

Now, we can get into a long drawn out discussion as to whether a bacteria in any form in the gut of an animal could be construed as a residue. But if you hark back to 512(d)(2), we are really looking at the effects of the use of a drug, and how those effects might be translated to the food.

We also, if we find a problem, we would like to see if we could mitigate this.


So why ask about pathogen load?

Well, there is a lot of data presented today, which, I think if -- well, there is a tremendous amount of data, and one interpretation of that data would suggest that maybe we did not quite get it right with 558.15 in the salmonella shedding studies.

I think that to take that a step further, and to attempt a meta-analysis of all of the data, both public and proprietary, that is out there, and to draw conclusions would be extremely problematic.

When we think about the safety of a drug product from the human food safety public health perspective, in the last couple of years we have come to asking a couple of fundamental questions.

This is actually something akin to a HACCP analysis. We want to specifically identify the hazard associated with a particular drug/bug combination; also, with respect to the production environment, the species, and the proposed conditions of use.

Because, for those not familiar with the system, we do not approve a drug product. We approve a particular proposed condition of use of a drug product, so that the condition of use is important, the production environment becomes very important.

So some questions on any of these microbiological issues that would apply to one species in one production environment with one drug just may not be applicable to another situation. And it really drives the experimental construct that you might come with to answer the critical questions to almost a case-by-case basis.

Now, I know that this was spoken to earlier today, and that we need to have some universal standard. I think we do need to have universal principles with respect the evaluation of the drug. I do not know that we could ever approach a universal experimental construct that would be appropriate to all situations.

So the question arises first, what do we really care about?

Well, we have heard from the FSIS folks about slaughter contamination and control of that contamination. We have heard from the production specialists about various levels of prevalence of infection on farm and in animals during transport.

I think that really if we focus down on the issue what we would care about is the bugs that are making it to the food commodity. And, yes, the meat is sterile when it is cut. But it is become infected, or it is becoming contaminated in the plant.

Now, one of your questions to the committee is, can you draw any kind of a line between drug use and this contamination?

And I would suggest that, in one sense, you can because that contamination is coming from the gut of the animal. I would also suggest that maybe the intercellular infection is only important to the extent that the intestinal mucosa ends up also being a contaminant in the slaughter house.

The FDA does not have the ability to control or prescribe evisceration techniques. That is not in our domain. But we do have the ability to control the events leading up to what happened to that if a drug is used in that process.

And so, I think it is important to consider that there may be production situations in which a drug is used, either therapeutically, prophylactically, or as a growth stimulant, in which the animal may not suffer other events.

We have seen corporate farming emerge in this country where animals do not go on trucks. They go from pen to slaughter. They do have stress certainly. There is the animal behavior. But we do not necessarily have that disjunction of several days with transportation that is incurring in every production environment.

So the point that I am getting to here is that we have, in chemical drug approvals, we have in the past considered inherent production withdrawal periods, and incorporated those into our evaluations.

We have come to regret that in some instances. And I think that it is important to consider potential worse case, and then to evaluate how significant is that worse case with respect to any public health problem.

The second area where I think that there is a potential for a problem is in environmental recirculation. We have heard about fecal contaminants, and their ability to infect naive animals with salmonella.

And, Dr. Wages, I am not going to pick on poultry just to pick on poultry. But I think that there is a good body of evidence that suggests that whatever is in the litter in a broiler house, regardless of top dressing or any other treatments, that what is in the litter is a major factor in what ends up in the guts of the birds that follow into the house in the next round.

So, we may want to know, we may want to ask a question. In that case, we would want to know about shedding rates, not about necessarily about what is in the gut of the animal at the time they are slaughtered, but we might want to know about shedding rates. And we would want to know that coupled with other information.

You could speculate that, in this case, campylobacter, which is not particularly viable under any circumstances, probably is not going to be there. So, even though campylobacter might be a public health concern with respect to chicken, you could maybe wipe that off your list with respect to an environmental recirculation question.

So, what I am saying, what I am trying to describe here is an approach that might be taken with respect to focusing in on what questions need to be asked. It is not within my competence to address exactly the mechanisms by which those experimental constructs would come about. I think that there are plenty of experts here who could deal with that, and have already spoken to the issue.

Finally, I would like to go back to the Act, and the requirements that we have placed on all other drug approvals. We do not prejudge any of the assays, the experiments that we request of sponsors, as to whether they will reveal something that is good, bad, or indifferent.

We are asking for data, the principle being that the sponsor needs to demonstrate to the FDA that a particular proposed condition of use is safe. I would find it very difficult to make a universal assumption that all uses of all classes of antimicrobials would be safe with respect to pathogen load in the absence of any data.

I think that there is a fundamental need to describe the effects, probably most of which will be beneficial to the sponsor, but we need to make an effort to approximate the proposed conditions of use and provide data which would support a conclusion of a reasonable certainty of no harm. Thank you.

Any questions?

Questions and Answers

DR. WAGES: Did you just not, in some of your comments, try to persuade this committee how to address these questions?

DR. ROBINSON: I tried to provide a perspective that I think has not been heard. I am just making suggestions. I am not trying to persuade. But these are concepts that I have not heard, and I think that they are important to bring out.

I am very interested in your comments with respect to this issue. And I am very willing to accept if you propose that it is a non-issue.

DR. WAGES: I just found personally some of your comments inappropriate coming to this committee in the format it was made.

DR. ROBINSON: I am sorry.

DR. LANGSTON: Any questions?

(No response)

DR. LANGSTON: Thank you very much.

DR. SUNDLOF: All right. Now, here to wrap this whole thing up on pathogen load, the session that you have heard this afternoon, we have Dr. David White, who is with our Office of Research, but is more recently on detail within the Office of the Director, dealing with many of our antibiotic issues. So, Dave.

MS. SINDELAR: This is our real computer expert right here. Thank you, Dave.

DR. WHITE: Thank you, Aleta.

Summarize and Adjourn Discussion

by Dr. David White

DR. WHITE: First of all, I would like to thank everyone for spending the whole day here. I know we have kept you over, did not realize we would stimulate this much discussion along with VMAC. And it has been a very good discussion.

I would like to take this time as well to thank the speakers for coming here, for the VMAC, as well, for coming here, and for all of the attendees for taking time out of their busy schedules to come here and talk about this subject.

We all know time is a very important commodity, and there is no price on that, and I appreciate everyone coming for that.

My objective today at the end is to summarize and adjourn, so I am going to do this quickly. I do not want to keep you here longer than we need to be.


Let's get a little history on this issue with CVM. It has been around a long time, the pathogen load. As Dr. Gilbert mentioned in the beginning, the CFR 558.15 studies were developed in the mid-1970s. They were focused on sub-therapeutic uses, and there was a shedding and resistance component to the studies.

As Dr. Robinson alluded to, in 1998, guidance for industry 78 was put into play, and it looks at all uses of antimicrobials, so not just sub-therapeutic, but therapeutic as well. And there is two components: One is a resistance component, where you look at the rate and extent of resistance; the second is pathogen load, and they are different.

And the VMAC today is looking at the second part, the pathogen load component. We are not looking about the resistance component today. And, in 2000, also as Mark presented, or mentioned, we had a CVM meeting on preapproval studies in resistance and pathogen load.

The focus really was on resistance, but pathogen load was part of the meeting agenda. That is on our website, if you all need to access that. And I believe that some people requested transcripts, and I hope that you have received those. If not, let us know, and we can get those to you.


One thing that is important to remember is that consideration of the potential microbiological effects of antimicrobial new animal drugs has been identified as a significant component of the animal drug safety evaluation process.

Also, the lessons learned from the 558 studies, along with 20 years of advancements in scientific knowledge, are important factors for helping to develop appropriate methodologies for evaluating the relevant microbiological effects associated with the use of drugs in food producing animals. Those are just two things to put out front.


I just wanted to mention again the speakers we have had today and summarized their talks. I am not going to reiterate what they said, just kind of some of the major bullets.

Dr. Sundlof started off the meeting introducing the subject. We had Jeff Gilbert talking from CVM, talking about the history of the studies. We had Tom Shryock talking about the conduct of the studies from the industry perspective.

Jeff Gray, from USDA, who has been very involved in these types of studies, talked about the design considerations. We had Dr. Mike Goodman from Exponent, who actually wrote -- was involved in the Exponent review that is on our web page as well.

Dr. Scott McEwen, who was able to travel from Great White North down here, talked about epidemiological evidence related to pathogen load. Dick Isaacson, from the University of Minnesota, talked about the other confounding factors that may affect pathogen shedding in food animals.

Dr. Barbara Masters and Karen Hulebak, from USDA FSIS, we thank them for coming out and talking to the VMAC with regards to what PRH, pathogen reduction HACCP, means; and also Dr. Robinson explaining from his perspective on the human food safety evaluation.


So just some bullets to remind you. Dr. Gilbert presented the brief history. He mentioned study design and examined parameters. He talked about the integrity measurements, the study results, also identified problems at CVM recognized, but he also mentioned that it was based on the policy and regulation of the time which was the 1970's.


Dr. Shryock also presented industry experience with these studies. He also talked about the design interpretation, the results. He also identified problems and limitations, and he questioned the relevance of such studies.


Dr. Gray presented information regarding the studies, and he mentioned some ideas about the organisms used, again, the study design, measuring effects, and some confounding factors.


Dr. Goodman, as a CVM contractor with Exponent to conduct a review of the literature, and Dr. Goodman tried to summarize that, and I said it is located at this web page. As well, it should be in the notebooks if you want to access that as well. They looked at 33 literature databases, and tried to break it down into challenge, or observational, or observational studies, and realized that there was limited data on the subject.


Dr. McEwen presented an overview of the relevant epidemiological information related to food animal studies, and he talked about human studies, animal studies, food animal studies, and he talked about some more study results. These are things that you have already said. I was just trying to summarize for you all.


Dr. Isaacson presented information relevant to shedding of pathogen food animals. He talked about the study designs, bacterial detection, culture methodologies, confounding factors, and some take home points.


Dr. Barbara Masters and Dr. Karen Hulebak talked about the information with regards to pathogen reduction HACCP. They gave us a brief history. The regulatory requirements, the implementation, performance criteria, and performance standards related to salmonella.


Dr. Robinson presented information describing evaluation of a sponsor-generated data. He talked about relevance to public health, identify and mitigate potential adverse human health effects, and he talked about the safety of the drug product.


Here are some key components to end on. The CVM recognizes that scientific information in this area is limited, and acknowledges the concerns raised at the February 2000 public meeting. Those are on record in the transcripts.

In response to that, we did contract with Exponent to try to get a grasp on the literature looking at these studies. And we feel that today's VMAC represents our ongoing efforts to look at the issue of pathogen load, and to develop appropriate policy in this area.

And to the VMAC, we are seeking recommendations on the issue evaluating potential antimicrobial drug effects on pathogen load in food producing animals, as part of the new animal drug application process. So, it is important to remember these approval studies that we are looking at.


Lastly, for tomorrow morning, just to give you an idea what is up, or an heads up, we have an open public session in the morning, and that's a.m. We have the presentation of the questions to the VMAC, committee deliberations, and the meeting summary at the end.

And, lastly, I would like to thank the members of the working committee on pathogen load. And that is Bill Flynn, Pat McDermott, Charles Easton, Karen Lampey, Jeff Gilbert, Aleta Sindelar, Mark Robinson, and Burt Mitchell.

And I thank you for your time. I do not know if there anything else you would like to say.

DR. SUNDLOF: Thank you. Thank you for that excellent summary in such a brief time.

And, at this time, I will turn it over to the chairman of the VMAC to adjourn. And, from myself personally, I hope everybody has a relaxing evening. Enjoy the Washington area. And, Mr. Chairman.

DR. LANGSTON: Thank you very much. I will just make a comment relative to Dr. Robinson's presentation. I think it is important that we know what the FDA is thinking, but I also understand Dr. Wages comment as well.

And, remember, from the committee standpoint, it is the science and whatever FDA thinks in the past/present, it will have no bearing on what we decide. So I will just end on that. Yes?

DR. PARKHURST: Before our experts leave, I do not know whether you will be here again tomorrow or not, but could I ask one question?

And maybe either Dr. Gray or Dr. Isaacson would be, or anybody else would be willing to address this, are there some pathogen shedding levels that are more hazardous than others, or would that be a relevant question to ask and have answered at some point?

DR. ISAACSON: I thing that is a great question. I do not know that there is an answer to it. We need to establish what is a level that is appropriate to which you do not want to exceed. And it depends upon the specific pathogen that we are working with.

Salmonella, the infectious dose, depends upon again the specific strain, but could be in the 105, 106 range for humans; whereas, E.coli 0157:H7 is down around single organism, 5, 10, 20, or cells per infectious dose.

So I think those things are things that we ought to be looking at and trying to get some data to establish some levels.

DR. PARKHURST: So would it be fair to say that it is detrimental to the environment, but not necessarily human health at this point?

DR. ISAACSON: I do not know that I would say that, because I do not think the pathogens in and of themselves are detrimental to the environment, per se. They are certainly not hurting the animals.

And the only time that it is detrimental is when the dose is high enough, exceeds some threshold, and a person consumes it, that it becomes a problem. So I do not know that it is an environmental issue in that sense.

DR. PARKHURST: Would it be true that we could always get rid of it? Or is there some threshold where it makes it just -- the problem is prevalent and it is very hard to get rid of?

DR. ISAACSON: My feeling about an organism like salmonella is that it is ubiquitous. I think Jeff Gray said that as well. You can find it literally anywhere. The idea of being able to eliminate it, is probably not a realistic goal.

I think that what is a realistic goal is the containing it, so that the levels are maintained below a certain level. And I am talking about within an animal. We can talk about carcass contamination as well. And I think there you can do things to try to reduce the number of carcasses as well that are actually being contaminated.

But I think the ultimate question is the specific level, and I think there are things that we can do to reduce that. But if the question were, can we eliminate salmonella, I would answer no.

Because there is this intimate association over time and evolution between the mammalian hosts, as well as other hosts -- it does not have to be just mammalian -- and salmonella. They have co-evolved and there is a strong association between them. And I do not think that it is a realistic possibility that that bond will ever really be broken.

But, as I was showing you in my data, is that the levels, the concentrations of salmonella per sample is reasonably low. And so, if we can keep it so that the interaction is good, but that it is not excessive, then it would okay.

I have a hunch if we try to do things to eliminate salmonella, it goes back to the question of antibiotic resistance, is that we are going to create a whole nother problem.

And there have been some epidemiologic associations, for example, between the emergence in poultry of salmonella enteritidis, and the decline, the directed decline of salmonella pullorum, the pullorum eradication procedures, and that what happened is that when pullorum was eradicated that there was a new, there was an ecologic space that was opened for which enteritidis could enter in. That is why we are seeing the epidemic of enteritidis.

So, I mean, I think that the best solution is a control of the population size, but not an eradication, because as soon as you eradicate or attempt to, then something else happens. And if it is not a resistance type thing, then it is going to be the emergence of perhaps a worse strain.

DR. PARKHURST: Tell me why control is important.

DR. ISAACSON: Well, why is control is important?

I think control is important from the standpoint of keeping the infectious dose sufficiently low, so that the risk is reduced. I do not think we can eliminate risk entirely as a food borne pathogen, but we can reduce the risk of it entering into the food chain, so that the incidence of diseases is lower.

So, if you can control it -- and I think a lot of these things have to do with hygiene and animal health, and processing, and abuse post-processing, and such, temperature abuse, for example, that will be useful in controlling the population size, per se.

DR. HOLLAND: I do not want to turn this into an issue to debate. But if salmonella is ubiquitous and everywhere, then why isn't it a problem in humans? Are humans carriers then?

MS. SINDELAR: Is this directed to Dr. Isaacson?

DR. HOLLAND: Well, we have heard it a number of times today that salmonella is ubiquitous and it is everywhere. Then are human beings carriers?

DR. ISAACSON: It is a good question. And I think the answer to the question is, no one has looked to ask the question. You know, it is hard to find it in animals, although we know it is widely distributed. You find salmonella frequently.

But people who are in the diagnostic labs in hospitals, unless they are looking for diarrhea causing organisms, are not going to be looking for salmonella. So I do not know the answer to the question, "Are humans carriers?"

We know that there is the famous carrier, Typhoid Mary, who carried typhi, but that was a unique and different kind of situations. Whether there are gut carriers, we really do not have very much data to say one way or another. It is possible; it is possible that they do not.

DR. HOLLAND: One more provocative comment.

DR. SHRYOCK: If I could comment on that just a little bit. I think what Dr. Isaacson was getting at, is if we took a survey of everyone in this room, we cannot answer whether someone in this room is carrying salmonella subclinically.

However, we do know that in cases where people have been infected with a non-typhoid salmonella, that they can carry it for years. The same is true in a sub-clinical sense for E.coli 0157:H7.

We have cases where a fully virulent organism can be carried for months with an individual and not show no clinical signs. So I think directing to the question you asked, no one has looked, but there is evidence to say, yes.

DR. HOLLAND: One last provocative statement. Then, if it is ubiquitous, and it is in humans, then why are we here?

MS. SINDELAR: I would just like to make two comments: One, that all of the speakers will be available for tomorrow for further questions, as a reassurance; however, Dr. Goodman will not, and Barbara Masters will not. So, if there are questions, then we can make the telephone calls as necessary to answer any inquiries.

And, second, just for transcription purposes, I just wanted to note that, thank you, Dr. Isaacson. You were the Mac user here. And so, we did have copies made, and the VMAC members were provided the copy prior to presentation. All copies will be available tomorrow. So, thanks very much.

DR. LANGSTON: I think my brain is getting too tired to deal with these issues anymore. So, barring any other comments, or an objection to adjournment, we will adjourn and reconvene at 8:30 in the morning.

(Whereupon, the meeting was adjourned at 5:40 p.m.)