Rapid evolution of fluoroquinolone-resistant Escherichia coli in Nigeria is temporally associated with fluoroquinolone use

Chloroquine has been described as one of the most important drugs ever used to treat an infection. This cheap and effective drug was the primary antimalarial worldwide until resistance emerged in the 1970s [1]. Multiple alleles of a chloroquine resistance transporter gene, pfcrt, and a multidrug resistance transporter gene, pfmdr1, are associated with chloroquine resistance in Plasmodium falciparum. Resistant P. falciparum haplotypes that predominate in sub-Saharan Africa appear to possess a fitness disadvantage, suggesting that they are maintained by selective pressure from chloroquine, which is still heavily used [2, 3]. Davidson et al [4] recently proposed that, in addition to the selective pressure for chloroquine-resistant plasmodia, chloroquine's weak antibacterial activity confers sufficient selective advantage for fluoroquinolone resistant Escherichia coli (FQREC), which emerged in rural Guyana in the absence of antibacterial quinolone use. If this cheap antimalarial selects for fluoroquinolone-resistant bacteria, justification for reintroduction-by combining it with resistance-reversing chemicals [5] or by withdrawing it until susceptibility is returned and then reintroducing it in combination with other antimalarials [2]-could decrease because of the great and growing need for fluoroquinolones. However, little is known about resistance, drug consumption and selection in malaria-endemic Africa, where high levels of chloroquine and fluoroquinolone use now occurs concurrently.

The quinolone antibacterial nalidixic acid was originally derived from a byproduct of chloroquine synthesis and shares structural features with chloroquine [6]. Nalidixic acid and its more active derivatives, the fluoroquinolones, were rarely used in Africa before the late 1990s because of their high cost. More recently, bacterial resistance to cheaper antibacterials has necessitated widespread fluoroquinolone use. Patents protecting ciprofloxacin and perfloxacin have recently expired, allowing for cheaper generics to be disseminated widely. Fluoroquinolones are stable, orally administrable, and now affordable. In the last decade, fluoroquinolones have become first- and second-line antibacterials of choice for acute respiratory, enteric and urinary tract infections as well as serious systemic infections such as bacteremia [7, 8]. Fluoroquinolones are also employed in combination with other antimicrobials to treat multidrug resistant tuberculosis [9, 10]. Additionally, their use and misuse in the informal sector, by unsanctioned providers and through self medication, is commonplace in Nigeria and in many other parts of Africa [11].

Quinolones inhibit the activity of bacterial DNA gyrase and topoisomerase enzymes, which are essential for replication [12]. Single nucleotide polymorphisms in the quinolone resistance determining regions (QRDR) of gyrA and parC, two of the genes that encode DNA gyrase and topoisomerase IV respectively, can lead to conformational changes in these enzymes that prevent quinolones from binding but still preserve their enzymatic function [13]. In E. coli and related Gram-negative bacteria, DNA gyrase is the first target for fluoroquinolones. If gyrA has resistance-conferring mutations, the primary target of fluoroquinolones switches from DNA gyrase to topoisomerase IV [12, 14, 15]. Mutations in the QRDRs of gyrA and parC are the most commonly documented quinolone resistance mechanisms, but resistance is also known to be conferred by mutations in the second topoisomerase gene, parE [14]. Quinolone resistance can also be acquired horizontally through transferable aac(6')-Ib or quinolone resistance (qnr) DNA. AAC(6')-Ib encodes a ciprofloxacin acetylating enzyme, while the product of qnr inhibits quinolones binding to target proteins [16, 17]. Another mechanism of quinolone resistance relies on upregulation of efflux pumps that export quinolones and other antimicrobials out of the bacterial cell. For example, mutations in the gene acrR, which encodes a repressor of the acrAB pump genes, are associated with quinolone resistance [18] Recently, qepA and oqxAB, which encode horizontally-transmitted efflux pumps have also been described [17, 19].

Low-level resistance to the quinolones often evolves by acquisition of one resistance-conferring mutation or gene. Subsequent genetic changes lead to higher levels of resistance to the first generation quinolone nalidixic acid and a broadening of the resistance spectrum to include second-generation quinolones (first generation fluoroquinolones) such as ciprofloxacin, followed by newer second- and third-generation fluoroquinolones that are yet to reach markets in Nigeria and most other African countries [12, 20]. Although multiple mechanisms of quinolone resistance have been reported from other continents, there are very few data from sub-Saharan Africa on the molecular basis for quinolone resistance. Prior to 2009, horizontally-acquired resistance had not been reported. If cross-selection from chloroquine results in antibacterial resistance more globally, then 1) FQREC should be commonplace in equatorial Africa, which has seen the greatest pressure from chloroquine, and 2) FQREC would have evolved before the introduction of quinolone antibacterials. We report quinolone susceptibility trends among commensal E. coli isolated from apparently healthy undergraduate students over three decades and antimicrobial use at the University health center over that period. Quinolone resistance emerged within the last decade and was coincident with the introduction of the fluoroquinolone ciprofloxacin.

This study allowed us to broach the almost untestable question of population-level selective forces for quinolone resistance in bacteria within a malaria-endemic area. The long duration of resistance monitoring offered significant advantages in spite of unavoidable limitations. We deemed it important to replicate the methodology of earlier years, which means that we only examined a small sample of the commensal E. coli population carried by each volunteer. Other limitations of this study include the fact that our prescription audit did not sample the complete catalogue of selective pressure for resistance. It is unlikely that patients who receive free medications at the health center would have chosen to purchase their prescriptions elsewhere, however, we cannot rule out the possibility that patients may have purchased other medicines to augment, or replace, those offered at the health center. Prescriptions for parenteral drugs are handled differently and records are not retained long-term at the center and therefore injectables were not evaluated, even though it is well known that many commonly used antimicrobials are administered intramuscularly in Nigeria [37].

The quinolone resistance rates and their association with access to quinolones that we have recorded in this study compare with those from reports from other parts of Nigeria and West Africa, and all support the hypothesis that fluoroquinolone introduction immediately precedes the evolution of FQREC [38–42]. However, it is difficult to draw even cautious conclusions across multiple studies and or from anecdotes. Therefore, in spite of the aforementioned limitations, the present study provides coherent evidence to support the hypothesis because it tracks resistance and use in a select area using simple but reproducible methodology. Our data show that quinolone resistance emerged in commensal E. coli after the 1990s in Ile-Ife, Nigeria, and that our 2005 isolate collection, largely but not entirely comprised of low-level resistors, appears to have been procured at a critical juncture-shortly after the evolution of resistance. By 2009, 16 (46%) of quinolone-resistant isolates showed high-level quinolone resistance, and resistance to the fluoroquinolones, attributable to quinolone-specific resistance mechanisms that have commonly been reported in other studies. Resistance-conferring mutations in gyrA and parC were the most commonly recorded of these resistance mechanisms. When horizontally-transmitted resistance evolved, qnrS1 was the most frequently detected transmissible allele among commensal E. coli from Vietnam and Ghana, where like Nigeria antimicrobials are freely available and fecal-oral transmission of bacteria is common [35, 43]. Altogether, we observed rapid, stepwise evolution of FQREC between 2005 and 2009, from a very low baseline.

The predominance of ST10 among 2005 quinolone resistant isolates may be due to the overall predominance of ECOR phylogenetic group A strains in West Africa [44]. However, we have recently shown that this lineage is overrepresented among quinolone-resistant strains, as compared to sensitive strains in Ghana [35]. We performed in-depth analyses of the 2005 isolates from this study to get a sense of how resistance might have evolved and our data suggest that strains with overactive efflux may have been the predecessors of strains exhibiting quinolone-specific resistance mechanisms. The differences in efflux phenotypes, efflux genes transcription profiles and flagellin types among most of the strains belonging to the same ST suggest that the low-level resistance we detect is not primarily due to expansion of recently derived clones. Instead, the data appear to suggest that elevated efflux is more likely to be seen in specific lineages, likely those with overactive efflux pumps. This in turn points to the idea that certain lineages may be primed to evolve more stable, less costly resistance. Horizontally-transmitted qnr genes, including qnrS1 identified in this study have recently been implicated in providing low-level quinolone resistance that could provide strains with the opportunity to acquire subsequent mutations in quinolone targets and therefore higher-level stable resistance [45]. It is therefore worrisome that qnrS1 can now be commonly detected among E. coli commensals in Nigeria. More studies are needed to improve our understanding of the evolution and dissemination of quinolone resistance. We suggest that on-going and proposed studies tracking resistance across Africa include a focus on quinolone resistance at genetic as well as phenotypic levels and that more work is done to elucidate the genetic context of successful horizontally transmitted quinolone resistance genes as they appear in African countries, similar to work that has been done in Vietnam [43].

Rampant non-prescription use, informal supply chains and a dearth of archival records make it challenging to quantify drug use in Nigeria. However, the health center at the University where we tracked resistance offers students and staff free health care and archives dispensed prescriptions. This long-term study was thus conducted in a setting that provided the rare opportunity to retrospectively study antimicrobial prescription patterns in an equatorial Africa locality where resistance was also tracked. We find that chloroquine prescription was inordinately high in the 1980s with 97% of patients visiting the health center receiving this drug. Chloroquine prescription declined steadily thereafter but never dropped below 500 daily-defined doses, with 25% of patients receiving this drug in 2006 (Figure 2). In contrast to the earlier report from Guyana [4], we do not find evidence of selection for FQREC from heavy and long-running chloroquine use in this setting. By contrast, quinolone antibacterials were never used before the 2000s and introduction of ciprofloxacin is temporally associated with evolution, and rising rates, of FQREC carriage by campus undergraduates. The significance of our findings cannot be over-emphasized since current efforts to 'bring back' chloroquine, an extremely effective and cheap antimalalarial, would be de-prioritized if there were adverse consequences on resistance in other organisms.

The number of drugs prescribed per patient and the common use of antimicrobial drugs we documented compare with studies performed elsewhere in Nigeria for the respective periods [46, 47]. Overall antimicrobial use was characterized by heavy dependence on three drug classes or less and a shift in favored drug categories from one decade to another, without withdrawal of earlier favorites. This drug use pattern, typified by the antimalarial chloroquine, favors the emergence of stable resistance [40, 48, 49].

Quinolones are drugs of last resort that are increasingly employed in Nigeria and other African countries to manage common bacterial infections. In the face of high and rising resistance to more affordable antimicrobials and the unavailability of newer, patent protected antibacterials, there is currently no real alternative to their use in West Africa. Evolution of resistance to the quinolones among E. coli has important clinical implications because resistance patterns in commensals typically mirror those seen in pathogens exposed to similar selection. A recent report from neighboring Cameroon suggests that quinolone resistance has emerged among Salmonella enterica serovar Typhi there [50] and we have reported quinolone resistance in epidemic Vibrio cholerae from Ghana [51].

This study charts the appearance of quinolone-resistant E. coli in a Nigerian malaria endemic community, which has seen heavy and long-term use of chloroquine. The study demonstrates that quinolone resistance was not temporally associated with chlrooquine use but appeared shortly after fluoroquinolone introduction. The diversity of quinolone-specific resistance mechanisms that emerged following ciprofloxacin introduction points to an urgent need for therapeutic alternatives and resistance containment. Our findings suggest that the useful lifespan for newly introduced antimicrobials that have been used extensively elsewhere can be short. High prices of patent-protected drugs delay their introduction to developing countries and strategies for ensuring early but controlled access to novel antimicrobials may be necessary to ensure medium- to long-term efficacy in resource-limited settings.