• Decrease font size
  • Return font size to normal
  • Increase font size
U.S. Department of Health and Human Services

Animal & Veterinary

  • Print
  • Share
  • E-mail

Generation, Presentation, and Application of Antimicrobial Susceptibility Test Data for Bacteria of Animal Origin: An Update From the Clinical and Laboratory Standards Institute (CLSI) by Jeffrey Watts, Ph.D., RM(NRCM), M(ASCP)

DR. WATTS: So, thank you very, very much, Pat for having me here today. And as Pat indicated I will provide you with an update of an ongoing activity at CLSI, which is a generation of a report entitled the Generation, Presentation and Application of Antimicrobial Susceptibility Test Data for Bacteria of Animal Origin. As I said, this is an ongoing activity and with a targeted publication date of late this year.


So, one of the things we have to think about whenever you talk to someone from CLSI is to understand the nomenclature that CLSI uses around documents. And so they’re basically three types of documents, there are standards, guidelines and reports.

And the standard is a consensus document that clearly identifies the specific, essential requirements for materials, methods or practices for voluntary use in an unmodified form. It is really is an SOP. And the standard may have discretionary elements included in it, however, it clearly defines as a discretionary elements. So it’s one of the things that people get themselves in trouble with is that they tend to use the standard more as a guideline rather than as a true standard. But it is to be used in an unmodified form.

Guidelines, however, are also generated through the consensus process, they describe criteria for a general operating practice. And the guideline can be used as written or it can be modified to fit specific user needs.

Reports, which I’m going to be talking about today, is a technical document that is intended for informational use only. It is not subject to consensus review. Reports in my parlance are the sticks in the sand. They basically establish -- they’re like a larger view article that establishes the state of the art at the time. They can be -- they can exist as stand alone documents, to be updated periodically, or what they can -- or what usually what happens is many times they’re used as a springboard to develop a specific standard or guideline.


And this gives you an example of the two standards or the standard and guideline that is mostly used by the VAST, M31, which is the performance standard for doing the actual susceptibility testing and contains the interpretative criteria, the veterinary specific interpretative criteria while M37 is the guideline in which we ask sponsors to bring data forward to us for setting those interpretative criteria and quality control parameters for their specific agent.


So, CLSI clinical breakpoints, as you probably are aware we’re primarily focused on the target animal pathogens and they have been focused on the need of the clinical laboratory to provide results predictive of therapeutic success. To set clinical breakpoints it requires what in the CSLI parlance a tripartite dataset including the microbiological distribution data, the PKPD data and then clinical efficacy data.

And those all have to be integrated to arrive at a single or single set of interpretative criteria. There is no set way of doing this. There is an algorithm in, M37 that will help you drive towards setting a breakpoint, however, it is not an absolute algorithm. There is no formula method for setting breakpoints.

And the results are reported as susceptible, intermediate and resistant. So, that brings that brings us to another section here. What do we mean when we say susceptible, intermediate and resistant, according to CLSI.


And we use these criteria for the definitions of susceptible, intermediate and resistant.

Susceptible is a category that implies that an infection due to the isolate may be appropriately treated with the dosage regimen of an antimicrobial agent recommended for that type of infection and infecting species, unless otherwise indicated. Generally speaking what it means is if you get a -- if you’re a clinician, you get a susceptible test result, that you will have a good likelihood of success if you choose to use that drug.

Intermediate, is the category that implies that you may be able to treat that infection in an area where the drug may be concentrated such as ß-lactamases and fluoroquinolones in urinary tract infections. But it is also quite often used as a buffer zone simply to prevent or reduce testing reporting errors due to variation in test methodologies.

Resistant isolates are not inhibited by the usually achieved concentration of the agent with the normal dosage schedules and/or fall in the range where there may be specific microbial resistance mechanisms and clinical efficacy has not been reliable in treatment studies. Another way of putting that when I talk to clinicians is that when you think of resistance think that you’re going to have decreased efficacy usually to the point that where you would not select to use that agent.


And so I want to give you an example here. And this is data that actually was originally published by Doctor Apley, Tilmicosin, but this shows you what a good, well-defined, clinical breakpoint does. And that is you have for Tilmicosin, you have the susceptible category of eight or less, you have intermediate 16, resistant at 32. When you look at the agreement between a treatment outcome and a susceptible isolate, that what you will find is, in this particular dataset, is the success rate for susceptible isolates is about 85 percent. And that is -- that is an agreement with what we commonly see with many human antimicrobials.

However, when we go to our resistant isolates, and I’m sorry, you can’t see this very well, when you go to the non-susceptible isolates, particularly the resistant isolates, efficacy drops to less than 40 percent. That’s a situation where you probably wouldn’t choose to use that drug. It is just not effacatious enough. This is what a good clinical breakpoint is designed to do.


The common errors that we see in using clinical breakpoints, one of the things that is commonly done is we fail to see or to maintain linkage between the method and the breakpoints. We consider that the breakpoints are specifically linked to the methodology, particularly if a methodology is -- if two methodologies don’t align very well in terms of their efficacy outcomes.

One of the things that we did early on in CLSI is we actually linked our breakpoints to the approved dosage and target pathogens only. So when you start going outside of those parameters you start breaking the linkage between the method and the breakpoint and it’s accuracy in able to predict clinical efficacy.

The only exception that you will see where we don’t use the approved dosage in indicated pathogens, particularly the approved dosage, is for the generic compounds, these are old compounds like penicillin, ampicillin and gentamicin, where over the years the dosage has crept up. And so now what happens is the generic working group looked at these compounds, they have established a breakpoint using what is, what they refer to as the standards of practice dose, as they exist currently today. However, when you go to table two, M31, and you look at the comment section, it will specifically give you the dosage that was used to set that breakpoint.


So, moving along here to XR-08, which is the report that we’re talking about, on the Generation, Presentation and Application of AST Data for Bacteria of Animal Origin, the document’s scope is to provide a review of current applications of susceptibility test data generated using CLSI methodology for bacteria of animal origins as well as recommendation for summarizing, presenting and applying the data in various ways.
More specifically the Report will provide an overview of the CLSI VAST approach to reference methodology, quality control and establishment and use of clinical breakpoints.

One of the things that XR-08 also will do is provide some guidance on areas where harmonization can be achieved among the various surveillance programs with the intent of being able to compare that data.

And so this is the subcommittee, and if you look at a CLSI subcommittee, one of the requirements that you have is you have to have representation by all the stakeholders. So you have to have sitting as committee members, you have to have industry, you have to have regulatory, and you have to have the academic and professions represented.

So we have Bob Adell which many of you may remember from his years with MicroScan, of course Patrick, Lori Moon, from the Animal Health Diagnostic Lab at Michigan State University, Doctor Stefan Schwarz from Germany, Doctor Al Sheldon formerly of FDA Cedar, Peter Silley, Doctor John Stelling from Brigham and Woman’s Hospital, and Bernd Stephan from Greater Animal Health.

Another person who has been quite active in this subcommittee has been Doctor John Turnage. And of course this is chaired by Doctor Simieei from Elanco.


So, why develop this document. Well, it’s been driven a lot by the needs of V-AST. That is one of the things that we find is that a development of clinical breakpoints requires a very large complex dataset. So a lot of information. Sometimes it can be difficult to obtain. And one of the areas, we have a working group on aquiculture that really struggled with this as the Aquiculture Working Group.

And so what they decided is that for many of their compounds in order to be able to provide some guidance to the laboratories for reporting out their results that they will use the epidemiological cut-off values and publish those.

Clinical breakpoints focus on predicting clinical outcomes. That’s one of the drawbacks of clinical breakpoints is they are not designed necessarily to detect early emergence of resistance. So that’s one of the things that you always have to remember, that clinical breakpoints don’t always do that very well. In some cases they do, but in many cases they don’t.

There is also the issue around the using the term susceptible and resistant. I think as microbiologist we tend to use the terms fairly loosely and we tend to understand what each other is talking about, however, whenever we go outside the microbiology community sometimes that clarity is lost.

And one of the things is the use of the wild-type or epidemiological cut-off values for reporting results in national surveillance programs.


So, the definition for -- this subcommittee is using for the epidemiological or “Wild-type” cut-off values, which I’m using, you will see this terminology, CO wild-type, which CO means cut-off. And that is the terminology that is used in algorithm for establishing M37.

But anyway the epidemiological cut-off value or CO wild-type separates the populations based on the MIC distribution by the presence or absence of resistance factors. The wild-types susceptible populations is assumed to show the antibiogram profile before any resistance has been developed or has been acquired or its distribution can be clearly differentiated from the resistant population.


And this shows you the contrast between clinical breakpoint and a epidemiological cut-off or wild-type. So if you look at this particular example where your wild-type cut-off of less than or equal to .25, this represents the wild-type population. Then you have a non-wild-type that includes the organisms at 1, 2 and 4. However, if you look at the clinical breakpoint because the dosage scheme that may be used for that particular compound allows you to successfully treat and your clinical data has established that you successfully treated isolates with MIC as high as four.

The clinical breakpoint is reflected as an MIC of susceptible breakpoint of less than or equal to four. While resistance is set up at greater than 64.


So, the pros and cons of using ECVs. The pros are that it will provide you with early detection of resistance development in given microbial population. They’re much easier to develop in clinical breakpoints. They don’t require quite the complex dataset. They can be developed for any agent regardless of clinical usage, which is a big advantage to them.

And the thing you have to understand, is wild-type cut-offs or epidemiological cut-offs are incorporated into all clinical breakpoints. They’re part and parcel of those clinical breakpoints.

The cons is they may not be predictive of clinical outcomes, they tend to in many cases be very conservative. They can be more sensitive to the sampling issues at times. And the methods for setting the ECVs are not completely standardized.


So, just to give you the working table of contents as they exist today, for XR-08 includes terminology, interpretation of the susceptibility test data, review of the current programs and what they’re using, analysis and presentation which includes some of the statistical analysis, some definitions around multi-drug resistance, and then some of the applications for various programs and the peer review process.


So, the recommendations that they’re driving toward right now is that the terminology should be standardized so that the breakpoint and I believe for the subcommittee is that at this point in time it would be clinical breakpoint, should be retained solely for clinical usage.

The epidemiological cut-off should be reserved to indicate reduce susceptibility, the wild-type population, as an epidemiological cut-off value for full susceptibility.

The key driver here and the key point I would like to leave you with is if we can come to adoption of some clear terminology it would enable us to provide the information clinicians use to treat individual animals, while also providing the information that the epidemiologist need in order to detect emerging resistance and to take the appropriate measures to control that resistance emergence.

So, with that, I believe that wraps me up and I thank you for your time and patience.


DR. McDERMOTT: Well, that finishes our session for this morning and brings us to our first official break. Thank you for enduring through for that long pause there.

We do have on the agenda now, for questions on the floor. So for any of the speakers, maybe it would be easier if the speakers use the floor microphones to answer the question and that would probably go a little bit more quickly that way.

As I said the important issue on the break is that the food and drinks I think disappear, according to our itinerary, so you can also continue discussions out in the hallway.