On the path to untreatable infections: colistin use in agriculture and the end of ‘last resort’ antibiotics

KEYWORDS:

• Colistin
• antibiotic resistance
• agriculture

1. Introduction

The unrelenting challenge of antibiotic resistance has led to the resurrection of classes of antibiotics that were supplanted by more potent and less toxic compounds, particularly for treating multidrug-resistant gram-negative infections. One example is colistin methanesulfonate sodium (CMS) a polymyxin class, ‘natural product’ antibiotic originally developed in the 1950s [1]. Soon after it was introduced into clinical practice, the detrimental effects of CMS on kidney and neurological function were recognized. CMS fell out of favor and its therapeutic niche was replaced by broad-spectrum β-lactam antibiotics, such as carbapenems. For many years, the carbapenems were regarded as drugs of ‘last resort’ and used to treat complicated, often hospital-acquired infections caused by pathogens like Pseudomonas aeruginosa, Klebsiella spp., and Acinetobacter spp. that demonstrated resistance to expanded-spectrum cephalosporins (e.g. extended-spectrum beta-lactamase producing Enterobacteriaceae and cephalosporin-resistant P. aeruginosa) [2]. Indeed, the rapid rise of carbapenem resistance in the past two decades has been the main driver for the resurrection of CMS [3,4]. The challenge faced by carbapenem-resistant Enterobacteriaceae and the increasing prevalence of imipenem-resistant P. aeruginosa were responsible for this practice.

But as evolutionary biology would predict, the increasing use of CMS and other polymyxins has led to the emergence of colistin-resistant bacterial strains. Most colistin resistance is induced by chromosomally mediated modification of the lipid A portion of lipopolysaccharide (LPS) in the outer membrane, leading to reduced affinity for polymyxins [5]. This type of resistance is common among different gram-negative bacteria pending unique variations and is not spread by horizontal gene transfer between bacteria. Other less common mechanisms also lead to polymyxin resistance.

Most recently, the world has been challenged by a novel resistance mechanism. Liu and colleagues [6] reported the in late-2015 plasmid-mediated colistin resistance by the mcr-1 gene (Figure 1). They found a high prevalence of mcr-1 in Escherichia coli isolates from retail meat, suggesting mcr-1 is widespread in food animals in southern China. Furthermore, their in vivo infection model showed mcr-1 provided adequate protection against colistin (minimum inhibitory concentrations [MICs] were raised to 6–8 µm/ml). Thus, the acquisition of mcr-1 by Enterobacteriaceae strains could make them resistant to all available therapies, if the mcr-1 gene was introduced in the correct background. Unfortunately, subsequent research confirmed mcr-1 is not confined to China and has spread across the globe [7–10]. This includes the United States, where mcr-1 was detected in a strain of E. coli cultured from a patient in Pennsylvania with a urinary tract infection and no recent international travel [11]. Retrospective studies have revealed that mcr-1 was present in isolates as early as the 1980s [12]. Xavier and colleagues analyzed bovine and porcine colistin-resistant E. coli isolates that did not carry the mcr-1 gene [13]. They discovered a novel plasmid-mediated colistin resistance gene, mcr-2 (Figure 1). Phylogenetic analysis determined mcr-2 likely originated in Moraxella spp., organisms that are intrinsically resistant to polymyxins. Of concern, the mcr-2 plasmid has a high transfer frequency and appears to lack a fitness burden on bacterial hosts.

Figure 1. mcr-1 and mcr-2 models as predicted by Raptor X [14] from Genbank entries AKF16168.1 & SBV31106.1 Coloring scheme corresponds to secondary structure and as visualized with Discovery Studio v3.1.

2. Why is it important to understand the mechanism of colistin resistance?

Because CMS was introduced in the 1950s and fell out of clinical favor about a decade later, it was only recently that our knowledge and understanding of the pharmacology and mechanisms of resistance to the drug came to equal that of other antibiotics. In general terms, resistance to CMS is acquired through LPS modifications that differ among gram-negative organisms. During CMS exposure, stable phenotypes caused by a pmrB mutation and non-stable colistin-resistant phenotypes are selected, suggesting that genetic mutations are unlikely to be the sole driver of CMS resistance [15]. This resistance may come at a cost to fitness and virulence. For example, colistin-resistant Acinetobacter baumannii isolates grow more slowly and are less capable of causing infection compared to nonresistant isolates [16]. Yet similar costs to biologic fitness have not been observed in other pathogens like P. aeruginosa and K. pneumoniae. One of the disconcerting findings from the report by Liu and colleagues was that the plasmid containing mcr-1 had a very high in-vitro transfer rate between E. coli strains [6]. Indeed, this plasmid is very stable and capable of transferring into epidemic strains of Enterobacteriaceae and P. aeruginosa without any apparent decline in fitness. Therefore, it seems likely that mcr-1 plasmids will be maintained in bacterial populations despite selection pressure from antibiotics. The mechanisms for the dissemination of this plasmid need to be elucidated in order to limit its spread into and between humans. Efforts to understand colistin resistance are also necessary to inform treatment strategies like optimal dosing regimens and combination therapy.

3. How is colistin resistance being spread by agricultural use?

Antibiotics have an important role in treating sick animals. However, most of the global usage is not for this indication, but instead goes for preventing infections and promoting growth [17]. The quantity of antibiotics given to animals is immense. For example, in the United States approximately 80% of all antibiotics used are fed to farm animals, amounting to more than 13 million kilograms annually [18]. Yet surprisingly few data are available about how these antibiotics are used. Furthermore, there are wide variances between countries in terms of total antibiotic usage and specific drugs employed. This may have more to do with economic and political considerations than medical ones. For example, in Denmark veterinarians are not paid by farmers, in contrast to the situation in Germany.

CMS is not used in agriculture in the United States but was the fifth most commonly used antimicrobial on farms in the European Union (EU) in 2011 [19]. CMS is administered to pigs, veal calves, and poultry to treat and prevent diarrhea caused by E. coli (called colibacillosis) and Salmonella spp. The amount of CMS used varies by EU member states, with Denmark considered to be a low user, France a moderate user and Germany a high user [20]. The reason for this discrepancy is unclear but different farming practices between countries is not a very plausible explanation. Moreover, it is somewhat surprising that France is not a low-user country given the successful public education campaign that reduced human antibiotic consumption about a decade ago [21].

Since China uses a vast amount of CMS in agriculture, it is likely selective pressure in the veterinary environment there led to E. coli acquiring the mcr-1 gene. CMS in animal feed probably led to a survival advantage for the strains carrying mcr-1 compared to the CMS-susceptible ones. The outbreak of mcr-1 containing E. coli of chicken origin started in 2009 and correlated with the increasing quantity of CMS used, ranging from 2470 metric tons in 2009 to 2875 metric tons in 2014 [12]. The effects of other antibiotics used in agriculture also seem to have a role in the spread of the mcr-1 gene, as multiple antibiotic resistance genes can be carried on the same plasmids or transposons. Investigators have determined the IncHI2 plasmid, which already carried resistance genes for quinolones (oqxAB), cephalosporins (bla_CTX-M-14), and fosfomycin (fosA3), recently captured mcr-1 and spread among different Enterobacteriaceae species on different continents [22]. Thus, in order to stop the dissemination of mcr-1 not only must CMS be restricted in agriculture but also other commonly used antibiotics.

4. What can be done to preserve colistin for human infections?

Fortunately, the issue of antibiotic use in agriculture and its impact on global antimicrobial resistance has been recognized by experts and authorities as a serious threat. For example, in 2014, the UK Government commissioned the Review on Antimicrobial Resistance in collaboration with the Wellcome Trust and the report was made public in May 2016 [23]. A key recommendation is the need for more surveillance data including the types and quantities of antibiotics that are being used in agriculture, as well as the emergence and spread of drug resistance in animals. One approach with great promise in molecular epidemiology is whole genome sequencing, which has been used to detect the mcr-1 gene in poultry [9]. This technology can be further advanced with primers and probes that target mcr-1 so that variants of the gene can be identified by a single sequencing reaction. Another example is GenEpid-J, an integrated database of genomics and epidemiology focused on plasmids that was launched in Japan in 2014. Japanese investigators assessed the prevalence of the mcr-1 gene in strains of E. coli collected from farm animals [24]. Of the 90 isolates with MICs to CMS of ≥8 mg/L collected between 2000 and 2014, only 2 were positive by PCR for mcr-1. While these results are enlightening, similar data are needed from other countries that use CMS in agriculture, especially China. Sustained funding for surveillance programs and to support collaborations between researchers investigating antimicrobial resistance in animals and humans is crucial. Recently, the UK government pledged 375 million USD through the Fleming Fund to bolster such efforts in low- and middle-income countries. Significant issues remain in the United States including a lack of any formal mechanism of surveillance for most antibiotic-resistant organisms on farms themselves, and the surveillance that is in place only looks for a few select pathogens from meat products.

The Review on Antibiotic Resistance recommended a ban on the use of antibiotics of ‘last resort’ in farm animals, including CMS [23]. However, the evidence is weak that such bans are beneficial and lead to fewer resistant bacteria in the environment. After a 2005 ban on fluoroquinolone use in poultry by the US FDA prompted by the clinical spread of ciprofloxacin-resistant Campylobacter jejuni, a decline in ciprofloxacin resistance was not observed [25,26]. Moreover, a ban by the EU on the use of the glycopeptide avoparcin in agriculture decreased vancomycin-resistant enterococci (VRE) in animals but did not lead to fewer VRE infections in humans [27]. One possible explanation is that not enough time had passed since the ban started for changes in antibiotic resistance trends to manifest. Conversely, it is also possible that resistant strains persist in the environment in the absence of antibiotic selective pressure due to the minimal cost to the organism associated with resistance [28]. Thus, the greatest benefit from a ban on CMS might not come from reversing resistance but instead from preventing further increases in prevalence and reducing the possibility that mcr-1 will be horizontally transferred into other bacteria. One potential barrier to a ban is the lack of veterinary oversight of antibiotics in agriculture present in many low-income countries [23]. Improving this capacity will take time but is essential for achieving long-term goals.

Another potential strategy is to develop new animal vaccines that prevent infections caused by drug-resistant pathogens. Vaccines are appealing because they can potentially break the cycle whereby a new antibiotic is developed and is immediately followed by the development of resistance by the targeted bacteria. Currently, there are many vaccines available for animal use [29]. The key to developing a vaccine against gram-negative pathogens that harbor colistin-resistance genes is identifying a lead antigenic target for active and passive immunization. These antigens should be conserved across clinical isolates and not homologous to the human or animal proteome. For example, investigators used a rational screening mechanism to detect an effective vaccine immunogen against A. baumannii, a pathogen that is becoming increasingly resistant to CMS [30]. Their analysis found that OmpA, a protein component of the outer cell membrane, induced protective antibodies in mice when given as passive immunization against lethal A. baumannii infection. Besides vaccines, other approaches to improve animal health are needed such better hygiene and reorganizing production sites to reduce disease. These are important considerations that require further scientific investigation.

5. Conclusions

History will remember the twentieth century as the time when antibiotics were discovered; hopefully, the twenty-first will be remembered as the time they were preserved and antibiotic resistance was tackled. Given the dire threat the spread of the mcr-1 and mcr-2 genes pose and despite a paucity of strong evidence, we believe a complete ban on CMS use in agriculture is reasonable. Enactment of a CMS ban will require engagement and buy-in from a wide range of sectors including governments, the pharmaceutical industry, farmers, veterinarians, and medical societies. Despite several previous calls for a ban on CMS use in agriculture [4,12,31], so far it has not been implemented on a wide scale. Inertia on this issue will lead to the continued dissemination of the mcr-1 and mcr-2 genes, resulting in patients with untreatable infections. Specifically, we urge the governments of China and the EU member states to restrict CMS use for human medicine only. Time is running out and we must act now.

Declaration of interest

RR Watkins has received grants for research from Allergan. RA Bonomo has received grant support from Astra Zeneca, Melinda, Steris, National Institutes of Health and the Harrington Foundation. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Acknowledgment

The authors thank Magdalena A. Taracila, MS, Laura J. Rojas, MS and Maria F. Mojica, MS, for designing the figure.

Additional information

Funding

This paper was not funded.