Bacterial resistance presents therapeutic dilemmas to clinicians worldwide. The warnings were there long ago, but too few people heeded them. Thus an emerging problem has grown to a crisis. Resistance is an ecological phenomenon stemming from the response of bacteria to the widespread use of antibiotics and their presence in the environment. While determining the consequences of inaction on the present and future public health, we must work to remedy the lack of action in the past. By improving antibiotic use and decreasing resistance gene frequency at the local levels, we can move towards reversing the resistance problem globally.
As we enter the new century, we finally see a consensus from various organizations and policy makers that antibiotic misuse is a major cause of bacterial drug-resistance, although the direct quantitative relationship between the amount of antibiotic used and the frequency of resistance is still lacking. This point may seem simplistic, but from the earliest signs of resistance, there was a reluctance among providers and manufacturers to link the problem to the misuse of antibiotics. Some considered clinical resistance an inevitable consequence of use. I disagree. Resistance becomes a problem when it presents in a resistant bacterial infection, not in an isolated bacterium. The clinical problem emerges when susceptible strains are decimated, permitting the resistant microbial flora to flourish and, in some environments, to gain prominence. These altered flora, a consequence of misuse and overuse of these valued therapeutics, are the source of resistant clinical disease strains. Bacterial resistance, which began an emerging problem, has grown to be a crisis .
In surveying the problem of resistance, we should focus on two components of the drug resistance phenomenon: the antibiotic agent and the resistance gene. Both are needed to produce a clinical resistance problem . In 1998, some 50 million pounds of antibiotics were produced in the United States, of which about half went to people in hospitals and homes. Of the remainder, about 80% was given to animals for various indications. The rest was applied to other organisms—honeybees, plants, trees, etc. In the early 1990s, I determined that 50,000 pounds of antibiotics were being used in agriculture annually under the designation “pesticides” . In the United States, pesticides include antibiotics, such as tetracycline and streptomycin. The Environmental Protection Agency (EPA) recently reported that 300,000 pounds of antibiotics were being sprayed onto fruit trees annually in the southern parts of the United States . This practice is also common in many parts of Central and South America. One can imagine the geographic spread of antibiotics by means of this application, complicated further by the dilution of the drugs as rain and other natural movements disperse them into the environment. The end result is excellent conditions for the selection of drug resistance.
Although the antibiotic agent is an important factor impacting antibiotic resistance, its selective role relates not only to the total amount, but how it is being used. The same antibiotic provided in different ways can have significantly different effects on antibiotic resistance. One study found that, when penicillin was given in less than therapeutic doses and for relatively long periods of time (5 days), patients' risk for carriage of penicillin-resistant pneumococci escalated . When the drug was used at the correct dosage for short periods of time, the emergence of penicillin resistant Streptococcus pneumoniae was considerably less . We can extrapolate these findings from the pneumococcus to many other organisms under selection pressure because of the use of antibiotic agents. It is low-level, prolonged usage that optimally selects for bacterial resistance. If we can change this practice, we can have great impact on resistance.
Furthermore, how an amount of an antibiotic agent is being distributed among individuals within a particular geographic area influences the frequency of antibiotic resistant bacteria. Giving 1000 doses of the antibiotic to 1 individual will have considerably less ecological effect on resistance emergence than giving those same 1000 doses to 1000 individuals. This concept is what I call selection density (figure 1) . We should focus not only on the amount of antibiotic, but also on how many individuals receive that drug and how confined is the area under treatment. It is a question of ecology—the net effect of the antibiotic on the numbers of resistant bacteria, and more importantly, on the residual surviving susceptible organisms left in the environment. The greater the numbers of susceptible microorganisms, the sooner the return to a susceptible microflora.
The other unknown and unstudied feature of the antibiotic is its “life after treatment” (figure 2). It may be that resistance emerges more readily in environments where the drug is deposited after drug application than it does in the actual patient, plant, or animal being treated. Antibiotics are very stable. Recently they have been detected in municipal waters . In my own laboratory, we stored tetracycline for several months in natural soils. When we extracted the soil, the tetracycline was still present in its original concentration (unpublished data). There are a relatively small number of mechanisms of resistance that actually attack the drug (table 1). β-lactamases destroy penicillins and cephalosporins, acetylation attacks chloramphenicol, enzymes inactivate the macrolides, and other agents inactivate the aminoglycosides. But most of the instances of emergence of antibiotic resistance are linked to a change in the target of the drug or a change in its transport . The latter mechanisms leave the antibiotic intact to continue the selection process. This is a feature of antibiotic resistance that has not been studied or quantitated with respect to the emergence of resistance. I suggest that environmental antibiotic residues may be the major force in selection and propagation of resistant strains (figure 2).
The second component impacting the resistance phenomenon is the gene itself. Resistance genes, carried on transportable elements, such as plasmids and transposons, can spread readily in the environment. Some time ago, workers showed that the transfer of a temperature-sensitive, transferable, multidrug-resistant H2 plasmid among Escherichia coli in cows occurred in the environment, not in the animals, whose body temperature did not permit the transfer . Outside the animal, the lower ambient temperature allowed transfer of the plasmids to new E. coli recipients. These E. coli were subsequently ingested by the cows, who then became colonized with the newly resistant strains. Such acquisition of new recipient strains did not occur in muzzled cows who were not able to ingest them from the environment. This was a convincing demonstration that transfer events occur in the environment.
Resistance originates as a local phenomenon but can expand to global proportions. Any localized problem can be transported, because international travel brings isolates from one country to another. In the 1960s, the spread of resistant bacteria among countries was demonstrated by J. Olarte , who reexamined some of the enteric pathogens in his stored collection. He found strains of Shigella dysenteriae that had resistance to chloramphenicol, streptomycin, and tetracycline, conferred in each case by the same plasmid, had been isolated in many different cities in South America (Figure 3) .
Genes are transferred among bacteria of vastly different kinds. On an historic note, it is interesting to reflect on comments by Sir Richard Sykes and Sir Mark Richmond over 30 years ago: “In practical terms, the intergeneric gene transfer of the type described here may mean that many genes, particularly those conferring resistance to antibiotics, that are usually encountered among the Enterobacteriaceae, may also occur among strains of Pseudomonas aeruginosa, and vice versa” . Our knowledge has developed significantly since these early observations. In fact, gene transfer can occur among gram-positive and gram-negative bacteria, such that a resistance determinant occurring in any organism can eventually reach any other organism. Although a resistance determinant may be transferred directly, it more likely follows a circuitous route, passing through multiple donors and recipients before taking up residence in a new genus of bacteria .
Yet another consequence of our inaction against uncontrolled antibiotic usage is manifested by multidrug-resistant bacteria excreted in human and animal feces. In a study we performed in the 1970s and 1980s, we found resistance to 2, 3, and 4 different antibiotics among the commensal fecal flora of human volunteers not ingesting an antibiotic. In those subjects consuming an antibiotic, multiple drug resistance (50% of the coliform flora) was significantly higher . Other investigators have reported similar findings [14, 15]. The new Web site on Reservoirs of Antibiotic Resistance (ROAR) maintained by the Alliance for the Prudent Use for Antibiotics (http://www.who.int/emc/WHONET/WHONET.html) shows that the environment is filled with ROAR genes in bacteria that may cause no disease but can be donors of resistance to potentially pathogenic bacteria.
Where did resistance in the fecal flora arise? Data suggest that it comes from the uncooked food we eat. In an early study, vegetarians were found to carry somewhat more, not less, resistant fecal flora than did meat-eaters . Although the meat might be contaminated with antibiotic-resistant bacteria, meat consumers generally cook the meat before eating it. Vegetarians eat uncooked foods. We found many lactose-fermenting organisms with multidrug resistance on carrots, celery, lettuce, cucumbers, peppers, and tomatoes . Denis Corpet examined 6 volunteers consuming a normal diet who were then switched to sterilized foods. There was a 1000-fold decrease in tetracycline-resistant bacteria with the sterilized diet . This study has at least 2 implications: food either carries resistant bacteria or it harbors residues of antibiotics. The finding gives us a reason for optimism, however. Our carriage of resistant fecal flora is not something we are destined to perpetuate. We need to clean up our food supply.
The real impact of this ubiquitous resistance is manifested in the hospital environment, where the major kinds and numbers of resistant pathogens have increased over the past decade. More recently, alongside methicillin-resistant Staphylococcus aureus (MRSA), MRSA with heterogeneous resistance to vancomycin have appeared. These can develop into strains we now call vancomycin-resistant S. aureus (VRSA) . We have vancomycin-resistant enterococci (VRE). Among the gram-negative bacteria, there are Klebsiella, Enterobacter species with extended spectrum β-lactamases, P. aeruginosa, and the newest member, Acinetobacter baumannii (table 2).
Does heterogeneous resistance to vancomycin have a clinical significance? In a group of 19 orthopedic patients infected with MRSA, vancomycin treatment failed in 13 . These were not people undergoing chemotherapy or who were otherwise immunocompromised. When the clinicians reviewed the patients, they found that the majority of the patients for whom treatment failed (12 of 13) were infected with MRSA that had heterogeneous vancomycin resistance . This is a highly significant difference and illustrates how even a small proportion of the bacterial population with resistance can impact clinical outcome. A recent paper reported that quinolone therapy failed for people infected with Salmonella typhimurium DT104 that had low-level quinolone resistance . Although not in the range of clinical resistance, the reduced susceptibility of the strains to fluoroquinolones appeared to have impact on the treatment outcome.
When Janice Bates at Oxford found the VanA determinant in enterococci that were isolated from farm animals , it took a while before the finding was linked to the use of another glycopeptide, namely avoparcin, as a growth promoter. There is still controversy over whether these animal strains are the reservoir for the strains that infect people. Of note, the vancomycin resistance determinant (VanA) did not stay in the E. faecium and E. faecalis, where it was first identified, but has been transferred to other species and other genera (table 3). The VanB determinant has moved into Streptococcus bovis . Transfer of VanA from Enterococcus to S. aureus has been observed both on filters and on mouse skin .
Our continued failure to contain the spread of resistance genes has now compromised health treatment in the community. We now have 6 major organisms that are problematic because of multidrug resistance (table 4). Even strains of Streptococcus pyogenes—a species that has historically been drug susceptible—have emerged that have macrolide and tetracycline resistance. To this list of organisms, we can add the strains of S. typhi with multidrug-resistance that have caused epidemics in developing countries .
Resistance has ecological consequences, some of which cannot be predicted. When second- and third-generation cephalosporins were introduced in the early 1980s for treatment of gram-negative infections, the enterococcus emerged as a new opportunistic pathogen—because of intrinsic resistance to the β-lactams. When we began to use vancomycin for MRSA, the first vancomycin resistance did not appear among the staphylococci but among the enterococci, which had been previously selected by the cephalosporins. Only now are we seeing vancomycin resistance in the staphylococcus . Imipenem used for Klebsiella, P. aeruginosa, Xanthomonas, and Enterobacter led to problems with Acinetobacter. We cannot necessarily predict the resistance outcome. It may or may not occur in the organism against which we are directing the treatment.
Trying to obtain information on the cost of resistance is difficult because there are few reported data. One study estimated costs between $150 million and $30 billion annually, depending on how many deaths were caused by resistance . On reviewing a series of studies involving susceptible and resistant bacteria, another group concluded that resistance leads to double the rates of morbidity and mortality and hospitalization . A more recent study by the Lewin group showed that MRSA, which can represent 60% of the S. aureus acquired in hospitals, led to 630 deaths in New York City, and very high costs for a single organism . We can clearly say resistance is costly. But, we need data on costs to convince government authorities to put more funds into confronting this problem.
There has been action, but unfortunately too little. One of the early examples that brought understanding and change in relation to antibiotics and the problem of antibiotic resistance in hospitals came from Henry Isenberg . He estimated that 15–30 liters of highly concentrated antibiotics were being sprayed into a 300-bed hospital environment yearly as a consequence of clearing the last air bubble from the syringe . His findings were instrumental in ending this practice. We now have new kinds of syringes, that do not require such preinjection maneuvering. We need many other such actions, both on small and large scales.
Inaction is one thing. Inappropriate action makes matters worse. The sale of household products with antibacterial agents has skyrocketed. There are now over 300 products that contain antibacterial agents on the nonprescription retail market today, from disinfectants to mattresses, in comparison to less than 40 such products 6 years ago. In Boston's Chinatown, you can buy antibacterial-containing chopsticks. Triclosan, a chemical used in hospital products, is now introduced into homes via hundreds of household items. We have demonstrated that triclosan has a target, the enzyme enoyl reductase, that is involved in fatty acid biosynthesis . Moreover, a genetic locus in E. coli provides resistance to triclosan and to multiple antibiotics via a multidrug efflux pump . Thus, there is a link between resistance to antibacterials and resistance to antibiotics. Selection of resistance by one group of agents can select for resistance to others.
During 1997–1999, there were 4 deaths from community-acquired MRSA infection among 200 individuals bearing MRSA in Minnesota and North Dakota . Strains have been isolated in Toronto and in other parts of the United States and in Europe. These MRSA did not originate from within the hospital. They are unique in bearing resistance only to the β-lactam antibiotics. There are many unanswered questions attached to this development Why are we facing MRSA in the community? What is the selective factor? Is it antibiotic use? Or are we converting our homes into hospitals because hospital-style antibacterial chemicals are now being used in the home? Presently we have no useful data for answering these important questions.
To a large extent, inaction characterizes our past, but it should not describe the present or the future. We need to act on the knowledge we have already gained regarding antibiotic resistance. One international organization, the Alliance for the Prudent Use of Antibiotics (http://www.apua.org), established in 1981, is dedicated to making a change through education and communication. It was difficult to mobilize action because few accepted that the problem existed. Fortunately, awareness has increased recently, but behavioral change is difficult to achieve. I hope that corrective action will come out of this meeting, by providing ways and directions to curtail and reverse the resistance problem we face today.
I thank Bonnie Marshall for her helpful comments in the preparation of this manuscript and in the design of figure 2.