Introduction

Local patterns of both human and agricultural antibiotic consumption have been shown to correlate with local rates of antimicrobial drug resistance for many antimicrobial-pathogen combinations [1], [2]. Thus, in truly remote communities, with little or no access to antimicrobials and limited interaction with other populations, rates of antimicrobial resistance should be negligible. High rates of fluoroquinolone resistance among Gram-negative bacilli (GNB) have been documented in tropical environments that have extensive use of fluoroquinolones [3]–[5], and have spread from there to the developed world [6]. To our knowledge, widespread fluoroquinolone resistance has not been previously documented to occur in the absence of fluoroquinolone use.

Chloroquine, 7-chloro-4-(4-diethylamino-1methylbutylamino)-quinoline, is extensively used throughout the tropics for its activity against erythrocytic forms of Plasmodium species. In Guyana, fever with no obvious anatomic source is empirically treated with chloroquine until malaria smears have been completed. In most malarious countries, chloroquine is still an important component in the treatment of P. vivax [7], [8], and despite resistance in P. falciparum [8], recurrent empiric treatment with chloroquine several times a year is widespread in the tropics, primarily due to its low cost [9]. In fact national household surveys in 28 African countries have shown that an average of 42% of children under 5 years with fever were treated with an antimalarial, and despite its poor efficacy in African children with P. falciparum, more than 80% of these reported treatments were with chloroquine [10]. Chloroquine is also used in the developed world in the treatment of rheumatoid arthritis, and systemic lupus erythematosis [11], [12]. Chloroquine inhibits DNA and RNA polymerase activity in vitro [13], halts the replication of Legionella pneumophila by limiting the availability of iron necessary for intracellular growth [14], and has weak antibacterial properties [15] against Salmonella sp. [16] and Bacillus subtilis [17].

The pharmacore of the fluoroquinolones and chloroquine are similar (Figure 1). In fact the origins of the quinolone class are from the use of chloroquine as an antimalarial [18]. A compound isolated from the commercial preparation of chloroquine was modified to produce the first marketed quinolone, nalidixic acid [18], [19]. Fluorine was subsequently added to produce the fluoroquinolones, resulting in both an increase in potency and spectrum. We hypothesized that the extensive use of chloroquine in tropical countries could select for fluoroquinolone resistance among E. coli despite the absence of fluoroquinolone use.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. A comparison of the chemical structures of ciprofloxacin (A) and chloroquine (B).

https://doi.org/10.1371/journal.pone.0002727.g001

Results

Surveillance cultures

A total of 535 rectal swabs were obtained: 240 in 2002, 152 in 2003 and 143 in 2005 (Table 1). Only 15.4% of individuals reported antibiotic exposure in the previous five years. No patients received antibiotics within 30 days of sampling. All participants denied fluoroquinolone use. In 2003, 30% reported chloroquine use within the previous 6 months and in 2005, 86% reported chloroquine use within the previous 2 years.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographics of the study population.

https://doi.org/10.1371/journal.pone.0002727.t001

A total of 30 ciprofloxacin-resistant E. coli isolates were isolated from specimens of 29 (5.4%) of the 535 participants (Table 1 and 2) (one patient in 2005 carried 2 different isolates), and were from children (12 of 225 [5.3%]) and adults (17 of 310 [5.5%]) with an age range of 8 months to 65 years. Resistant isolates were obtained throughout the study period [8 of 240 (3.8%) in 2002, 16 of 152 (10.2%) in 2003, and 5 of 143 (3.5%) in 2005.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of ciprofloxacin-resistant E. coli isolated from Guyana.

https://doi.org/10.1371/journal.pone.0002727.t002

The ciprofloxacin MIC ranged between 2 and >32 µg/mL, with 26 of 30 E. coli having MIC of >32 µg/mL (Table 2). The chloroquine MIC ranged between 5,000 and 80,000 µg/mL. By contrast, the chloroquine MIC of susceptible E. coli isolated from Guyana and Canada ranged between 625 and 1,200 µg/mL. PFGE typing identified 24 distinct strains among the 30 isolates (Table 2).

All ciprofloxacin resistant isolates had mutations in parC and gyrA commonly associated with fluoroquinolone resistance (Table 2). No qnrA/B genes and no evidence of an active efflux system were identified. No QRDR mutations were found among ciprofloxacin-susceptible E. coli isolated from the same patient population.

Chloroquine selection of fluoroquinolone resistance in vitro

After serial exposure to chloroquine, three isolates were able to grow in the presence of ciprofloxacin at 0.032 µg/mL (Table 3). The ciprofloxacin MIC increased from 0.008 to 0.094 µg/mL in one isolate and to 0.12 µg/mL in two isolates. Sequencing of the QRDR regions of gyrA and parC revealed an aspartate-87-tyrosine mutation in gyrA. Serial exposure to increasing concentrations of chloroquine resulted in a further increase in the ciprofloxacin MIC from 0.12 to 6 µg/mL in two of the three strains. Analysis of parC and gyrA revealed a second mutation in the QRDR of parC resulting in a change from serine-80-isoleucine. No mutations in gyrB or parE were detected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. In-vitro selection of ciprofloxacin resistance after exposure to chloroquine.

https://doi.org/10.1371/journal.pone.0002727.t003

Discussion

To our knowledge this is the first report implicating quinoline antimalarial use in the development of bacterial quinolone resistance in the tropics. The resistance mutations that were found in the Guyanese samples, and selected in vitro by chloroquine exposure, result in cross resistance to all quinolone antibiotics (including nalidixic acid, norfloxacin, ciprofloxacin, levofloxacin, and moxifloxacin).

Antibiotic resistance has become a global crisis as a result of widespread use of antibiotics in medical and agricultural practice [3]–[5]. However, in the remote Amerindian communities of Guyana there is limited access to antimicrobials. Moreover, there are geographical barriers limiting travel in and out of this region. The villages are located in the Guyana highlands, surrounded by the Pakaraima Mountains and dense tropical rainforest. The villages themselves are at significant elevation (Kamarang is at 1621 ft). One therefore requires air travel to enter this area (143.8 mile flight from Bartica), and then boat transport between communities. These are the headwaters of the Mazzaruni River, and so contamination of water from higher up in the river does not occur. This geographic isolation and thus difficulty with travel leads to financial barriers preventing inhabitants from venturing beyond the region, and also preventing outsiders from venturing in. Furthermore as reserved Amerindian communities, there are political limitations preventing outsiders from travelling into the region [20]–[22]. Whether these communities will be able to remain separate and maintain their culture is unclear due to the fact that mining companies are now hoping to enter this area [23]. Due to the isolation of these communities during the course of our study however, it would be expected that the Amerindians would be subject to negligible rates of antimicrobial resistance. Despite the lack of quinolone exposure, we found faecal carriage of E. coli resistant to ciprofloxacin in 5.4% of this population. A large intensive care unit survey in the USA done at the same time as our study, found only 4% of E. coli resistant to ciprofloxacin [24].

We cannot with certainty dismiss the possibility that a few individuals traveling outside of the region may have been exposed to fluoroquinolones, and on return, transmitted resistant organisms to their neighbours. However, this hypothesis is unlikely not only because of the isolation of these communities, and thus the rarity of outside travel, but also because the resistant E. coli were primarily polyclonal and the strains were widely distributed among inhabitants of different villages. The fact that 24 different clones of quinolone resistant E. coli were found is evidence that an occasional traveler would be an unlikely source of the multiple resistant strains and therefore, there must be a local factor selecting for resistance.

Detailed demographic data on the involved communities, and regarding the patients presenting to the clinics, and those tested, are presented in table 1, as well as Appendix S1, Appendix S2, Appendix S3 and Appendix S4. The data suggests that the most common reason for presentation to the clinics was feeling well but wishing a check-up (35%). Fifty three per cent presented with one of the three most common reasons (check-up, musculoskeletal pains, or request for family planning). In fact 82% of patients did not have any infective diagnosis, and only 16% had a bacterial etiology suspected. Furthermore, none of the patients in our study had received an antibacterial antibiotic within 30 days of presentation to the clinic. Therefore we believe that our data reflect the prevalence of quinolone resistant E. coli in the general population.

Chloroquine is a weakly active antibacterial with a long half life (4 days) that is given frequently and excreted through the bowel; all factors that may optimize the opportunities to select for bacterial resistance. In this study, laboratory experiments were able to demonstrate that chloroquine exposure in-vitro can select for QRDR mutations in E. coli. Unlike quinine, chloroquine is not felt to have a natural counterpart. We suspect that chloroquine found in the drinking water in Kamarang may be the result of contamination by human waste. This suggests widespread exposure to chloroquine, even in the absence of individual malaria treatment. It is also notable that chloroquine use within two years of testing was almost universal and thus identification of an association between chloroquine use and ciprofloxacin resistance within individuals was not possible. Nevertheless, resistance in E. coli was highest in the villages that had recently experienced a dramatic P. vivax outbreak, and the corresponding increase in chloroquine use may have led to an increased prevalence of ciprofloxacin-resistant E. coli. Specific data regarding total chloroquine use in these communities is not available. However we inferred a dramatic increase in use in late 2002, as chloroquine is used for the initial treatment of all cases of suspect malaria until smears are available, and is then continued for confirmed cases P. vivax. Therefore the 1,270% rise in P. vivax and the 916% increase in total malaria cases during the epidemic period of late 2002, implies a dramatic increase in chloroquine use at that time. It is notable that this data consists of the reported incidence of smear positive and thus TREATED cases, (not including estimates of undiagnosed cases) and therefore does reflect actual chloroquine use.

Chloroquine levels were measured in the water in a non-epidemic year (2005) and thus may have been higher at other times. Although the levels we detected are unlikely to select for fluoroquinolone resistance, detection of chloroquine in the river, likely as a result of human waste, may be an indirect measure of the high volume of chloroquine use in the region. As traditional dosing for P. vivax used by the medics includes chloroquine 1 g PO then 500 mg PO 6 hours later, and then 500 mg PO daily×2 more days, very high concentrations are achieved in the upper gut, and these concentrations would exceed those which were used in vitro to generate resistance. A gradient of concentrations would likely be generated throughout the intestines and this may facilitate the emergence of resistance. It is possible that the presence of small concentrations of chloroquine in the drinking water may have provided an ongoing selective advantage to ciprofloxacin resistant E. coli within patient's intestines, prolonging carriage.

Some local residents may have acquired quinolone resistant E. coli from exposure to contaminated drinking water. Nevertheless, the identification of 24 distinct isolates of quinolone resistant E. coli in this population, suggests that a single point source of water contamination would not be able to explain the multiple circulating strains. Rather generation of resistant strains is likely occurring in many different patients' intestines, with some horizontal transmission within the community via drinking water contamination.

The potential exposure to both quinolone resistant organisms and chloroquine itself from human waste contaminated river water may explain the failure to identify an association between individual reported use of chloroquine and carriage of quinolone resistant E. coli. Furthermore the reliability of individual reports of chloroquine exposure is unknown. Nevertheless the fact that increased total community chloroquine use (as inferred by testing in a P. vivax epidemic year versus non-epidemic years) was significantly associated with prevalence of resistance, is strong ecological evidence of the association of quinoline use with the development of quinolone resistance in the community. This situation is analogous to that seen with antibacterial antibiotics, where there are very few reports of individual use being associated with the development of resistance, but the evidence of total community use leading to increased resistance is more compelling [1], [2]. Indeed, this may be one circumstance in which ecological data are more useful than individual-level data.

Although no clinical outcomes were impacted in this area by quinolone resistance, (due to the lack of local quinolone use), we believe the same phenomenon is likely also occurring in other malarious communities, and may be leading to a lack of clinical efficacy of quinolones. This site was particularly interesting to study, because the lack of the confounding presence of quinolones in the community allowed us to identify the relationship between community wide quinoline use for malaria and quinolone resistance in bacteria.

One limitation of this study is that we failed to rule out the possibility that other quinoline antimalarials, particularly primaquine, which together with chloroquine is the standard treatment regimen in Guyana for cases of P. vivax, may also be selecting for ciprofloxacin resistance. Other quinoline and quinoline-like compounds include quinine, mefloquine, quinidine, halofantrine, lumefantrine, piperaquine, pyronaridine, and amodiaquine. Other study limitations include a limited number and timing of water samples, and failure to ask about chloroquine exposure in the first year.

This study is the first to identify human carriage of quinolone resistant organisms in communities that lack any quinolone exposure. It is also the first to identify antimalarial therapy as a factor in the development of carriage of quinolone resistant bacteria. We have reported the selection of E. coli resistance by chloroquine in this study. A recent report demonstrated that fluoroquinolone resistance may result in an increase in mortality from E. coli infections due to the increased likelihood of inadequate empirical antimicrobial therapy [25]. A similar phenomenon likely also occurs in other organisms treated empirically with quinolones. Indeed persons in these very remote Amerindian Guyanese communities were found to carry multiple other quinolone resistant gram negative organisms including Salmonella (data not shown). Quinolone antibiotics have an important role in the therapy of many tropical infectious diseases such as typhoid fever, non-typhoidal Salmonella, Shigella and enterotoxigenic E. coli infections. They also are important in both the tropics and in temperate regions in the treatment of gonorrhoea, urinary tract infections, upper and lower respiratory tract infections, are becoming increasingly important in the therapy of tuberculosis, and have widespread use in hospitalized patients. These antibiotics have excellent oral bioavailability, and do not require refrigeration, making them a critical component of our antibacterial armamentarium in the tropics, and for these reasons the development of resistance to these agents has important consequences. As chloroquine use is widespread throughout the tropics our findings may represent an important public health problem. In fact, although P. falciparum chloroquine resistance is widespread, use of quinoline antimalarials will likely continue, and may increase, as they are used in artemisinin combination therapy (ACT) [26]. The large scale recent introduction of these therapies, especially as presently contemplated for use in mass intermittent presumptive therapy (IPT) for population wide malaria control [26], could have a significant impact on the antibacterial efficacy of quinolones.

In light of the large number of lives saved through proper malaria treatment/control, our data does not imply that quinolines should be discontinued from use. However we plan to carry out further studies is to identify whether some quinolines may be less likely to induce quinolone resistance than others, and thus be safer for malaria control programs. Our data also implies that further investigations regarding the efficacy of quinolone antibiotics in other regions of high malarial endemicity are necessary, particularly since many of these areas presently lack facilities for bacterial culture and sensitivity testing, and thus rely on empiric treatment. Furthermore our data adds impetus to programs that lead to the primary prevention of malaria, such as vector control, and the development of vaccine programs, in order to reduce the need for widespread quinoline chemotherapy.

Methods

Supporting Information

Acknowledgments

We would like to thank the Ministry of Health of Guyana, Ve'Ahavta: the Canadian Jewish Humanitarian and Relief Committee (and in particular Sarah Zelcer, Leya Aronson and Bekkie Vineberg), The Lions Clubs of Guyana, the medics and doctors of Guyana region 7, and the many villagers who volunteered for this study.