April 9, 2001

Dockets Management Branch (HFA-305)
Food and Drug Administration
5630 Fishers Lane
Room 1061
Rockville, MD 20852

Re: Docket #00D-1677

Dear Sir or Madam,

The undersigned health, environmental, consumer and agricultural groups submit the following comments on FDA’s document titled "An Approach for Establishing Thresholds in Association with the Use of Antimicrobial Drugs in Food-Producing Animals" (Dec. 19, 2000), hereafter referred to as the Thresholds Document.

We commend the Food and Drug Administration’s Center for Veterinary Medicine (FDA-CVM) for taking steps to address the use of antimicrobial drugs (hereafter referred to as "antibiotics") in food animals. The FDA-CVM recognizes the impacts this practice has on public health by promoting the spread of antibiotic-resistant microorganisms. Antibiotic resistance is a complex issue of critical importance and we applaud the Agency for making it a top priority.

In the Thresholds Document, FDA lays out a conceptual approach for addressing resistance issues in considering applications for approval of new antibiotics for use in food animals. Specifically, whenever FDA approves a new agricultural antibiotic that is also used for treating foodborne illness in humans, it would set two "thresholds" – one based on levels of resistance to the antibiotic in bacteria isolated from animal-derived food and a second one based on levels of resistance in human isolates. If either threshold were crossed, FDA would initiate proceedings to withdraw the approval. The thresholds would be set at levels designed to prevent adverse health consequences to humans.

While a thresholds approach is appealing in concept, current limitations in scientific understanding raise real issues of whether it is now possible to set thresholds that will be protective. This is particularly so for antibiotics that are not now approved for use in humans, if they (or a close chemical relative) may be needed for use in humans in the future. As the virginiamycin/Synercid example teaches, antibiotics should not be used in agriculture in ways that limit the efficacy of vital human drugs at a time when the cupboard of effective human-use antibiotics is becoming increasingly bare.

The population dynamics of antibiotic resistance are highly complex and not well understood. The number of resistant organisms changes drastically as microbial populations expand and contract, and the hazard is not necessarily limited to a particular species of microorganism.

In practical terms, it may be possible to support a threshold-based approach, insofar as it demonstrably incorporates proven methods for detecting antibiotic resistance related to agricultural uses. In order to accomplish this, FDA-CVM must publish methods adequate to this goal and not define thresholds solely in terms of levels of bacterial resistance in human isolates. We praise FDA-CVM for recognizing that antibiotic-resistant genes identified in animal isolates will inevitably reach humans. It is important to recognize that, in addition to foodborne pathways, environmental-contamination pathways may also be important (e.g., via groundwater, surface water, soil contact, etc.).

Accordingly, in implementing a thresholds approach, FDA must take a number of considerations into account. These include:

• The potential length of withdrawal proceedings, during which time the product remains on the market (some prior withdrawals have taken years to complete);
• The limitations of current microbial-surveillance systems and the resulting difficulty in detecting exceedence of a threshold;
• Uncertainties (due to limited scientific understanding) as to whether any observable resistance threshold for animal isolates will actually suffice to protect human health;
• The large quantities of antibiotics used in food-producing animals that are not related to foodborne pathogens, particularly in light of the ability of unrelated bacteria to swap genes that code for resistance traits.

For these reasons and others, thresholds would need to be set at very stringent levels and surveillance must be very rigorous in order to achieve the "reasonable certainty of no harm" standard that governs new animal drug approvals. In many instances, the threshold may well be lower than the current limit of detection, meaning that FDA-CVM should refuse to approve that antibiotic for use in animal agriculture.

One further point: the Thresholds Document addresses the approval of medically important antibiotics for therapeutic and non-therapeutic uses in agriculture. Rather than developing thresholds for antibiotics that are (or may become) medically important, we urge FDA-CVM to state explicitly that it will no longer approve such drugs for non-therapeutic use in agriculture. It is simply not acceptable to jeopardize the efficacy of human medicines for, at best, slight efficiencies in animal production. For the same reason, FDA-CVM should move promptly to withdraw existing approvals of medically important antibiotics for non-therapeutic use.

If the Thresholds Document is revised to reflect the following list of concerns, however, it can become an appropriate tool to consider requests for approval of new therapeutic antibiotics for use in agriculture.

The Thresholds Document implies that development of resistance will be a linear process over time. In doing so, the Document fails to consider the fact that resistance follows logarithmic patterns in the presence of antibiotic pressure owing to the dynamics of microbial growth. The existing surveillance and monitoring techniques, based upon a linear assumption, cannot provide timely and accurate information on whether that point is being exceeded or whether the resistance is on the log growth phase.

The model also implies that upon removal of the antibiotic, the trend of resistance will simply be reversed. This is not always the case and depends upon many factors, such as other sources of pressure, common mechanisms of resistance, the costs of resistance, and the stable carriage rate of resistance genes in the particular bacteria. In the European Union, for example, the use of avoparcin (a glycopeptide similar to the human-analogue vancomycin) as a growth promoter in animals was banned in 1996. The levels of vancomycin-resistant enterococci (VRE) in humans in Germany decreased from 12% in 1994 to 3% in 1997. By contrast, 4 years after the prohibition of using tetracycline as a growth-promoter in hogs, there was no significant reduction in the resistance levels of E. coli in the hogs. Such uncertain dynamics in the loss of resistance make determining a static threshold for antibiotics challenging.

As alluded to in the previous paragraph, a threshold approach is only acceptable insofar as it can be detected and responded to. Existing monitoring systems would need to be greatly expanded in order to ensure that appropriate data is available to determine whether the thresholds have been crossed.

The current National Antimicrobial Resistance Monitoring System (NARMS) needs to be strengthened so that it is adequate to the task of detecting when resistance thresholds have been crossed. NARMS now consists of two branches. The first branch monitors the resistance in human isolates and is under the auspices of the Centers for Disease Control and Prevention (CDC). The second branch monitors the levels of antibiotic resistance in food animals and is run by the United States Department of Agriculture (USDA). Although this system provides a snapshot of national antibiotic resistance, it is still limited in its capabilities. The emergence of antibiotic resistance in foodborne bacteria will not likely affect all populations at an equal rate. As far back as two decades ago, scientists demonstrated that people living and working near a farm using antibiotics had increased levels of antibiotic-resistant bacteria in their intestinal flora. There are no systems to detect this type of transfer; CDC’s emerging infections network only follows hospital-based cases of infection rather than community based sampling of humans for colonization by resistant organisms. Sites that cover areas with higher densities of farms using antibiotics in livestock may demonstrate a problem before other sites in the country, and yet the overall figures may not show a significant trend. FDA-CVM must enhance NARMS’ coverage of both human populations (including farming communities) and food products to help bolster the data and strengthen the overall detection capabilities of the system.

The document primarily focuses on zoonotic bacteria as the pathway for the transfer of antibiotic resistance to humans, ignoring the multiple pathways that exist for human exposure. These other pathways include, but are not limited to, the spread of antibiotics and antibiotic-resistant bacteria to ground and surface water, the contamination of soils and crops from the spread of manure used as fertilizer, and the infection of farm workers and their families in contact with animals and agricultural antibiotics. There is also no discussion concerning the transfer of resistant genes to commensal organisms, a factor that has significant implications for human health. Because bacteria are ubiquitous and can share genes – particularly plasmid-borne genes, such as those encoding resistance traits – with other unrelated microorganisms, thresholds must be set conservatively. At the point at which exceedence of a threshold is detected, reservoirs of resistance have accumulated and spread in the environment far beyond the original source.

The threshold document focuses on new animal drug applications (NADA) for antibiotics but does not appear to address the large quantity of medically important antibiotics that are used in food animal production today. Estimates of the fraction of antibiotics used for food animals in agriculture range up to 80% of the total usage in the United States. Millions of pounds of penicillin, tetracycline, lincomycin, erythromycin, tylosin, bacitracin and virginiamycin have been given to farm animals for growth promotion and other non-therapeutic purposes for decades. These are antibiotics that are used, or are close analogues to antibiotics that are used, to treat important human infections. They are not, however, generally used to treat foodborne illnesses (with the exception of erythromycin, which is used in the treatment of human Campylobacter foodborne illness).

The extensive use of antibiotics promotes development of reservoirs of antibiotic-resistant bacteria in the environment, which may transfer their genetic material to human pathogens or, under certain circumstances, become pathogenic themselves. For example, Bacteroides can acquire tetracycline-resistance genes from commensal bacteria in the human gastrointestinal tract. Although Bacteroides are not harmful in the lower intestines of humans, they can cause life-threatening infections when they escape from the gut during surgery or following abdominal trauma. They are thought to be the most numerically dominant bacteria in the human large intestine and are the most important anaerobic bacteria associated with human infections. The greater the opportunity these bacteria have for exposure to antibiotics and/or to obtain antibiotic-resistant genes, the greater the threat to humans.

By focusing solely on the treatment of foodborne illnesses in humans, the Thresholds Document fails to address concerns about development of resistance to antibiotics used for other purposes, and about non-food pathways for spread of resistance. Moreover, by focusing on illness rather than colonization, it will be impossible to detect early signals of resistance transfer.

The Thresholds Document mentions, only in passing, that improved management practices may reduce the reliance on use of antimicrobial drugs in food-producing animals. Good husbandry practices have enormous impact on animal health and can do much to reduce the need for pharmaceutical solutions. Animal health is dependent on the level of exposure to and colonization by pathogenic organisms, individual animals’ ability to mount an immune response and, lastly, the availability of effective pharmaceuticals. Some features of many intensive animal production systems reduce the ability of livestock and poultry to mount a natural immune response when exposed to diseases.

A variety of U.S. producers now raise livestock and poultry successfully and profitably without routine administration of antimicrobials for growth promotion or for prophylactic means. Similarly, limitations on antimicrobial use for livestock and poultry in the European Union caused minimal or no impact on productivity, animal health or profitability. Increased attention to disease-resistance traits by livestock- and poultry-breeding companies, together with improvements in husbandry practices, would substantially decrease the need to rely on antimicrobials in livestock and poultry production.

We recognize that animal husbandry techniques are not FDA-CVM’s direct responsibility. Nevertheless, FDA-CVM could play a major leadership role in encouraging government and private sector research aimed at using animal husbandry techniques to reduce antibiotic use.

FDA’s withdrawal process for drugs can be extremely lengthy. The withdrawal of nitrofurans, which FDA initiated in 1971 was not completed until 1991. Similarly, FDA initiated proceedings to withdraw diethylstilbestrol (DES) in 1972, and completed the proceeding in 1979.

Because antibiotics can continue to be used until the time that FDA concludes a withdrawal proceeding, antibiotic resistance can continue to increase. FDA must set thresholds at levels low enough to take into account the possibility of a multi-year delay between the time the threshold is crossed and the time that the product can no longer be legally used. If FDA-CVM fails to do so, the threshold approach cannot possibly achieve the "reasonable certainty of no harm" standard that governs new animal drug approvals, and some antibiotics may be lost as effective human drugs.

We appreciate this opportunity to comment.

Very truly yours,