We reviewed and mapped antimicrobial resistance in aquatic food animals in Asia during a period of substantial industry growth. Our findings indicate that between 2000 and 2018, antimicrobial resistance in bacteria from cultured aquatic food animals was stable (33%) while the resistance from wild-caught aquatic food animals decreased sharply (52% to 22%). These trends represent currently available evidence from point prevalence surveys, which serve as a surrogate in the absence of systematic surveillance and should be interpreted cautiously. Structured, systematic surveillance will be imperative to document trends in multi-drug resistance at the sub-national level in the future.
Our results are consistent with an analysis of antimicrobial resistance in aquaculture-derived bacteria from forty countries, nearly half of which in Asia, which identified a global mean multi-antibiotic resistance index of .25, and a higher index (>.35) in low-income and middle-income countries in Asia27. Although antimicrobial use in surveys from cultured animals was most frequently unspecified, in the limited surveys that recorded whether on-farm antimicrobials were either used or not used (n = 63; 11%), use was associated with higher multi-drug resistance than the absence of use (p < .001) (Supplementary Fig. S8). The combined effect of multi-source antimicrobial introduction15 and persistence30,31 in aquaculture waters and sediments may present a consistent selective pressure facilitating the maintenance of an elevated level of antimicrobial resistance.
Declining resistance observed in bacteria from wild-caught aquatic animals could be associated with reduced human and livestock fecal pollution exposure in wild-caught animals over time. Anthropogenic influences such as human and livestock fecal pollution have been positively associated with antimicrobial resistance gene frequency32,33, and in these surveys, we observed a negative correlation between multi-drug resistance and the World Bank basic sanitation index34 (Supplementary Fig. S9). In addition, surveys from the first half of the study period held a significantly higher proportion of diseased animals than those from the latter half, and disease was positively associated with resistance (Supplementary Fig. S10). Importantly, these findings must be considered in the context of relatively scarce point prevalence surveys (11%, n = 81) attributable to capture fisheries. Other contributions—including improved wastewater management and regulatory action—cannot be ruled out and future studies will be valuable in documenting resistance trends and their influencing factors in capture fisheries.
Temporal trends in P50 rates reflect resistance across multiple bacteria, aquatic animal species, production contexts, and subregions in Asia. Production dynamics, antimicrobial exposure intensity, and regulatory standards, amongst other factors, may be highly variable by location and at timescales shorter than the duration of our analysis. Consequently, these trends may not capture more granular country-level resistance profiles in select pathogen species from specific aquatic animal production settings (Supplementary Fig. S11).
Amongst foodborne pathogens, we observed a concerning dual-threat profile, marked by high rates of resistance to first-line antimicrobial classes and, for Vibrio and Aeromonas spp., moderate to high rates of resistance to antimicrobial classes of last-resort reserved for treatment of multi-drug resistant pathogens.
In Western and Southern Asia, Aeromonas spp. isolates exhibit high rates of resistance to tetracyclines, sulfonamides, aminoglycosides, monobactams, carbapenems, and cephalosporins, which, together with moderate resistance to fluoroquinolones, suggests therapeutic options for management of invasive or non-self-limiting Aeromonas infections may already be restricted in this subregion as compared with lower resistance profiles reported outside of Asia35,36. High rates of carbapenem resistance from Aeromonas spp. were also identified in this subregion. Aeromonas spp. produce inducible metallo-beta-lactamases, are inherently resistant to ampicillin, amoxicillin-clavulanate, and cefazolin, and are known to express a carbapenemase encoding gene (cphA) the significance of which remains unknown37. Recent work identified the fish pathogen Aeromonas allosaccharophila as the origin of mobile cephalosporinase genes, which confer resistance to beta-lactams, and may be associated with aquaculture antimicrobial use38. While beta-lactam antimicrobials are utilized in aquaculture5,39, to the best of our knowledge, carbapenems are not used. Elevated carbapenem resistance identified in Western and Southern Asia may reflect inducible resistance under high environmental exposure to beta-lactam antimicrobials in this subregion, and, taken together with increasing carbapenem resistance in Aeromonas spp. recently reported from other regions40,41, may warrant further investigation.
Although variable by subregion and foodborne pathogen, high rates of resistance to third-generation and fourth-generation cephalosporins were also observed. These findings are consistent with trends identified in terrestrial animal production26. Ceftriaxone, Ceftazidime, and Cefotaxime resistance in E.coli across subregions was 27.1% (95% CI 25 to 29%), but approached 40% in Western and Southern Asia and exceeded 50% in South-eastern Asia, suggesting that resistance to third-generation cephalosporins is established in at least some members of Enterobacteriaceae in Asia. Notably, resistance in E.coli must be interpreted in the context of human and terrestrial animal fecal pollution and may represent the serial accumulation of antimicrobial resistance genes as has been observed in some human pathogens such as Shigella42. Extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae capable of hydrolyzing third-generation cephalosporins are recognized as a serious threat to human health43. Our finding confirms reports44,45 of an existing pool of ESBL genes that may be plasmid encoded46 and raises the prospect of horizontal gene transfer conferring broader resistance within and across diverse bacterial genera with the consequent loss of this critically important collection of cephalosporins.
Elevated colistin resistance in Gram-negative bacteria from aquatic animals observed in this study mirrored high rates identified in terrestrial animals in Asia over a similar study period26. Colistin resistance across a breadth of bacterial genera in Asia may reflect both intrinsic and acquired resistance mechanisms. Intrinsic resistance to polymyxins in Gram-negative bacteria has been documented47. The possibility that a proportion of the colistin resistance identified is acquired cannot be ruled out, particularly through plasmid-mediated transfer amongst members of the Enterobacteriaceae family48. Such resistance may be attributable to the prolific use of colistin and other polymyxins in the region, which is only recently evolving through national regulatory action, and the possible recruitment of mcr-family mobile resistance determinants, their broad dissemination through horizontal gene transfer, and the wide distribution potential of aquatic environments14,48,49. Indeed, the distribution mechanics of AMR in aquatic systems under differing hydrological conditions is an expanding area of research that holds the promise of contributing to an enhanced understanding of AMR risk across compartments50.
Previous work has shown that bacterial foodborne disease attributable to aquatic animal consumption contributes a non-negligible burden to human morbidity51,52. In this context, the high rates of resistance to multiple classes of critically important antimicrobials in aquatic animal foodborne pathogens in our study raise urgent concerns regarding both therapeutic efficacy of first-line antimicrobials and the further erosion of last-resort therapeutic options for multi-drug resistant infections resulting in severe disease.
Hotspots of AMR in freshwater environments were predicted along several of the region’s great river systems, including the Yangtze River, and the estuaries of the Mekong and Nile Rivers. Low rates of AMR were predicted in Japan and South Korea. In these countries, the predictions were associated with high uncertainty (Supplementary Fig. S18). In our freshwater data set, South Korea contributed 2.4% of surveys and there were no surveys from Japan. Consequently, predictions in these two countries should be interpreted cautiously, requiring validation through additional surveillance. However, low rates of AMR could indicate less influence from human and livestock fecal pollution. Low AMR may also reflect comparatively lower rates of integrated livestock-aquaculture farming, enhanced environmental pollution regulatory capacities, and a heightened awareness of antimicrobial stewardship principles in these countries. Low AMR in the peri-urban environment of Guangzhou in southern China was unexpected considering the comparatively dense human population, prevalence of integrated livestock-aquaculture farming systems, and broad agricultural and human connectivity with the aquatic environment in Guangdong Province14. Future work will be important to corroborate these findings and identify factors that have the potential to moderate multi-drug resistance.
We identify 50 hypothetical future survey locations prioritized to maximize the knowledge gained over the current map of resistance from freshwater in Asia. Weighting the uncertainty in predicted resistance by the product of human population density and inland aquaculture production prioritizes future surveillance effort to locations where potential multi-drug resistance exposure risk and impact on human health and the aquaculture industry may be most significant (Supplementary Note 4). China and India together account for 72% of future surveys, reflecting their comparatively high population density and predominant role globally—first and second, respectively—in inland aquaculture production2. Our approach is scalable to individual country contexts and presents an opportunity to fill national or sub-national knowledge gaps and inform interventions and stewardship under resource-limited settings by maximizing information gained through targeted surveillance.
In marine coastal environments, the highest rates of AMR were identified in northeastern China on the Yellow and East China Seas; southern China and central Vietnam on the South China Sea; southern India on the Arabian Sea and the Bay of Bengal between southern India and northern Sri Lanka; and the eastern Mediterranean Sea on the coast of Lebanon. Lower AMR rates were observed along Thailand and Malaysian coastal waters both on the Gulf of Thailand and the Andaman Sea. Aquaculture production has been correlated with aminoglycoside resistance gene abundance along estuarine waters of coastal China53. High rates of AMR in waters also identified as carrying the highest potential for marine aquaculture productivity54, such as the coastal waters of Hainan Island on the South China Sea, may help prioritize surveillance in an industry poised for substantial growth. However, marine predicted values should be interpreted with caution, given the comparatively limited availability of marine surveys overall (n = 322) and an absence of surveys from countries such as Indonesia with sizeable marine aquaculture and nearshore marine fisheries industries.
As with any modeling study, our analyses come with limitations. First, point prevalence surveys, although numerous in the study (n = 749), present an inherent challenge to standardization of data from susceptibility testing that are subject to variability in methodologies and protocols resulting in uncertainties55. Second, the resistance across species within bacterial genera can vary markedly. Our analyses of drug resistance were conducted at the bacterial genera level—the most granular level at which complete data were available. Although analysis of trends at the bacterial species level may be valuable in context-specific settings, the aim of our study was rather to document temporal trends over twenty years in Asia across a broad range of bacteria. Third, our study includes samples originating from both healthy and diseased animals. An over-representation of samples from diseased animals associated with treatment failure could present a bias toward higher resistance rates56. Indeed, samples from diseased animals had higher median P50 values than samples from healthy animals. However, any influence on resistance across the study is likely moderated by a balanced mix between healthy (42%) and diseased (58%) animal samples from the surveys and our treatment of intermediately susceptible isolates as susceptible in our analysis. As drug tolerance precedes clinical resistance57, it is possible that our results underestimate non-susceptibility. Fourth, the volume of data is currently insufficient to map AMR across different study periods using spatio-temporal interpolation methods. Overall, there was an increasing volume of surveys conducted in recent years (Supplementary Fig. S4). However, when evaluated by time series that most closely partitions each of the cultured and wild-caught data sets in half (cultured: 2000 to 2010 vs. 2011 to 2018; and wild-caught: 2000 to 2009 vs. 2010 to 2016), the directionality of the regressions remained unchanged, suggesting the increasing volume of surveys does not influence the temporal trends in P50. Fifth, due to the contiguous relationship between aquatic animals and their environmental waters, the interpretation of resistance in commensal isolates such as E.coli must be contextualized as indicative of human and livestock fecal contamination32, either in origin waters or through post-harvest contamination. While characterizing the origin of selective pressures driving resistance is critical to risk mitigation at source, our study documents elevated resistance in bacteria from aquatic animals intended for human consumption—irrespective of origin—and can thus be interpreted as a risk to human health from contact with or consumption of aquatic animals and their products. Finally, it is notable that no surveys were available from Indonesia, despite contributing nearly 10% of global aquatic animal production. Vietnam—the fourth largest global producer of cultured fish1—represented less than 3% of all surveys. An absence of surveys from Myanmar and Laos also illustrates gaps in understanding antimicrobial resistance in aquatic food animals in Asia. Despite our broad search parameters, surveys conducted in these countries and which were either not identified in our search or not available for review could potentially influence our findings. Similarly, in the absence of systematic surveillance, reliance on passive surveillance data presents variability in survey coverage, which could introduce bias, as well as uncertainty in survey geolocation, adding potential uncertainty to predictive models (Supplementary Fig. S18). Our study addresses this uneven geographic representation of point prevalence surveys by identifying—on an objective basis—sub-regions in Asia that would benefit most from further surveillance efforts. Such future surveillance will be essential to an enhanced and refined understanding of resistance trends in Asia.
This study identified elevated rates of antimicrobial resistance in bacteria isolated from aquatic animals intended for human consumption in Asia. A growing aquatic food animal production industry may serve as an important pathway for transmission of resistance along the food chain with potential consequences for human health.
A scale-up of an antimicrobial resistance surveillance architecture for aquatic food animals is urgently needed to fill gaps in AMR trends at national and sub-national levels. Our findings help direct the prioritization of this future surveillance effort and provide a foundation for establishing time-bound, measurable targets for reducing antimicrobial resistance58,59. The spatial profile of antimicrobial resistance presented here should inform planning for sustainable development of a high-growth aquaculture industry, critical to feeding an expanding global population54,60 while balancing the imperative for healthy freshwater and marine environments, and the preservation of antimicrobial efficacy for future generations.
Source: Ecology - nature.com
