Preliminary insights into the occurrence of similar clones of extended‐spectrum beta‐lactamase‐producing bacteria in humans, animals and the environment in Tanzania: A systematic review and meta‐analysis between 2005 and 2016

The emergence and spread of extended‐spectrum beta‐lactamase producing Enterobacteriaceae (ESBL‐PE) are complex and of the public health concern across the globe. This review aimed at assessing the ESBL‐PE clones circulating in humans, animals and the environment to provide evidence‐based insights for combating ESBL‐PE using One Health approach. Systematic search from Medline/PubMed, Google Scholar and African Journals Online was carried out and retrieved nine eligible articles (of 131) based on phenotypic and genotypic detection of ESBL‐PE between 2005 and 2016 in Tanzania. Analysis was performed using STATA 11.0 software to delineate the prevalence of ESBL‐PE, phenotypic resistance profiles and clones circulating in the three interfaces. The overall prevalence of ESBL‐PE in the three interfaces was 22.6% (95% CI: 21.1–24.2) with the predominance of Escherichia coli (E. coli) strains (51.6%). The majority of ESBL‐PE were resistant to the commonly used antimicrobials such as trimethoprim–sulfamethoxazole and tetracycline/doxycycline, 38%–55% were resistant to ciprofloxacin and all were sensitive to meropenem/imipenem. ESBL‐PE infections were more associated with deaths compared to non‐ESBL‐PE infections. Strikingly, E. coli ST38, ST131 and ST2852 were found to intersect variably across the three interfaces. The predominant allele, blaCTX‐M‐15, was found mostly in the conjugative IncF plasmids connoting transmission potential. The high prevalence of ESBL‐PE and shared clones across the three interfaces, including the global E. coli ST131 clone, indicates wide and inter‐compartmental spread that calls for One Health genomic‐driven studies to track the resistome flow.

agents as growth promoter as well as unregulated antimicrobial disposal to the environment have been pointed out as the predominant factors driving the persistence of AMR (El Salabi, Walsh, & Chouchani, 2013;Kummerer, 2003;Martinez, 2009;Wegener, 2003).
Moreover, coresistance with non-beta-lactam antibiotics such as aminoglycosides and fluoroquinolones has been reported to complicate antimicrobial therapeutical options for both enteric and extraintestinal infections (Pitout & Laupland, 2008;Storberg, 2014). The burden of ESBL-associated infections is growing with noticeable regional variations, and higher ESBL-PE rates have been reported in Asia, Middle East and Latin America, whereas the rates in North America and European countries have remained below 10%, with a few exceptions in some countries (Hoban et al., 2012;Morrissey et al., 2013;Sturenburg & Mack, 2003). Available limited local data from the African continent also show a continuous rise in ESBL-PE rates in most countries (Storberg, 2014). The CTX-M types of beta-lactamases, notably the bla CTX-M-15 , have been observed to be the most common ESBL alleles (Pitout & Laupland, 2008). It has recently emerged as a worldwide allele in isolates causing infections in the healthcare and community settings, but also colonizing animals and birds (Nicolas-Chanoine, Bertrand, & Madec, 2014).
The ESBL-PE burden among humans and animals has been widely described in Tanzania (Lupindu et al., 2014;Madoshi et al., 2016;Sonda et al., 2016), as opposed to their occurrence in the environmental sources (Lyimo, Buza, Subbiah, Smith, & Call, 2016). There is high faecal carriage of ESBL strains among children and in the general population, which in turn may partly explain their involvement in fatal sepsis among vulnerable children population in this country (Blomberg et al., 2005;Kayange, Kamugisha, Mwizamholya, Jeremiah, & Mshana, 2010;Mhada, Fredrick, Matee, & Massawe, 2012). Moreover, previous studies which evaluated AMR in the three interfaces were not specifically focused on ESBL-PE (Lupindu et al., 2015;Shah, Colquhoun, Nikuli, & Sorum, 2012). Therefore, there is limited information on the linkage between ESBL-PE in humans, animals and environmental interfaces in Tanzania. In the present review, the aim was to examine whether there is existence of similar clones of ESBL-PE in humans, animals and environmental interfaces in Tanzania. To address this, a review of available literature was carried out to establish the prevalence of ESBL-PE and ascertain whether there are similar ESBL clones circulating between humans, animals and the environment, as a crucial starting point for specific interventions.

| Literature search and selection criteria
Literature search of articles on ESBL-PE in humans, animals and the environment in Tanzania was carried out by two researchers (JS and NM) from Medline/PubMed, Google Scholar and African Journals Online (AJOL) from October to December 2016. Whenever there was a discordance finding, consensus was reached by involving a third researcher (SEM). Abstracts and titles of 131 articles published in English from 1946 to 2016 were retrieved using exact key words ("AMR" AND "Tanzania") as recommended by the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines (Table S1) (Moher, Liberati, Tetzlaff, Altman, & PRISMA Group 2009), and all articles on AMR searched have been attached (Table S2). We used Medline/PubMed search strategy to retrieve 72 abstracts and titles. This was followed by specific search using key words ("ESBL OR [extended-spectrum beta-lactamase]" AND ["Human" OR "Animal" OR "Environment"] AND ["Tanzania"]), which in turn enabled us to retrieve more 59 abstracts and titles (48 from Medline/PubMed and 11 from Google Scholar/AJOL). A total of 116 full-text articles were retrieved for in-depth assessment of eligibility.
Studies which included both phenotypic and genotypic ESBL detection from humans, animals and environmental sources were considered eligible for the final systematic review and meta-analysis.
Moreover, review articles, laboratory-based studies focusing only on resistance genes transfer, and studies conducted during outbreaks were excluded (Figure 1).
Articles were further assessed for quality and strength of evidence, and graded as high, middle or low based on criteria stipulated before with some minor modifications (Hedin & Källestål, 2004;Storberg, 2014) (Table 1). Therefore, articles included in the final analysis (n = 9) contained information on the prevalence of ESBL-PE, circulating ESBL clones, phenotypic resistance profiles markers and proportions of ESBL strains expressing bla CTX-M-15 (Table 1).

| Data extraction, quality assessment and data analysis
Information extracted from the articles was author names, year of publication, journal name, study duration, study site/region, study population (humans, animals and environment), types of samples,

Impacts
• The overall prevalence of extended-spectrum betalactamase producing Enterobacteriaceae in humans, animals and from environmental sources was 22.6%, with the predominance of Escherichia coli strains.
• Escherichia coli subtypes ST38, ST131 and ST2852 were found to intersect variably across the three interfaces.
• These findings underscore the need to institute One Health molecular-based approach for in-depth assessment of transmission dynamics in the three interfaces so as to guide specific preventive measures against these multi-drug-resistant bacteria. Two strategies were employed in assessing potential bias in eligible studies, and as a result, two studies (Doijad et al., 2015;Mshana, Gerwing, et al., 2011) were excluded because they were conducted during an outbreak involving one strain of Enterobacter spp. Also, five studies were excluded because they included only genotypic detection of ESBL (Lyimo et al., 2016;Madoshi et al., 2016;Doijad et al., 2015;Mshana, Gerwing, et al., 2011;Sato et al., 2009), inclusion of these studies would have resulted into involvement of ESBL strains containing ESBL genes which are not expressed phenotypically ( Figure 1). Nevertheless, contributions of these studies were appraised and discussed in this review.

| Predictors and outcomes of ESBL-associated infections/colonization in humans and animals
Of the nine articles included, four analysed factors associated with ESBL-PE infections/colonization (Blomberg et al., 2005;Mshana et al., 2016;Seni et al., 2016;Tellevik et al., 2016). In domestic animals, exotic breed type was significantly associated with ESBL carriage compared to local type whereas (Seni et al., 2016)

| Antimicrobial resistance patterns of ESBL bacterial isolates to non-beta-lactam agents
High-resistance trends of ESBL-PE to the non-beta-lactam antimicrobials were observed in all three interfaces (Blomberg et al., 2005;Moremi et al., 2016;Mshana et al., 2013Mshana et al., , 2016Ndugulile et al., 2005;Onken et al., 2015;Seni et al., 2016;Tellevik et al., 2016). In all cases, human isolates exhibited more resistance to the four antimicrobials compared to animals' and environmental isolates. The majority (60%) of ESBL-PE were resistant to the commonly used antimicrobials such as trimethoprim-sulfamethoxazole and tetracycline/doxycycline, 38%-55% were resistant to ciprofloxacin and between 46% and 75% were resistant to gentamicin.

| ESBL alleles/genes and clones circulating between humans, animals and the environment
All but one study (Onken et al., 2015) Seni et al., 2016). In humans, E. coli ST131 was the commonest followed by ST38  and ST617 predominated in animals (Seni et al., 2016). A quarter of the isolates expressed unique STs Mshana et al., 2016;Seni et al., 2016). Strikingly, E. coli ST131, ST38 and ST2852 were the only clones found to circulate in both three interfaces Mshana et al., 2016;Seni et al., 2016)

| DISCUSSION
In this review, the overall ESBL-PE prevalence pooled from the three interfaces was found to be 22.6%, which is within the range of ESBL prevalence in other African countries such as Kenya, Algeria and Cameroon irrespective of the categories (Kiiru, Kariuki, Goddeeris, & Butaye, 2012;Storberg, 2014). However, this overall prevalence is lower compared to 42% reported among humans in East African hospitals (Sonda et al., 2016). Prevalence in the later may be accounted by the fact that these ESBL bacterial isolates were from patients in the hospitals where MDR bacteria prevail. This is also supported by two studies in Uganda where the proportions of ESBL-producing bacteria isolated from inpatients and outpatients were over 75% and 5.3%, respectively (Najjuka, Kateete, Kajumbula, Joloba, & Essack, 2016;Seni et al., 2016).
As it has been reported in other studies done outside African continent (Friedmann et al., 2009;Pasricha et al., 2013), carriage of ESBL-PE in this review was predicted by history of previous use of antimicrobial agents and longer hospitalization (Blomberg et al., 2005;Mshana et al., 2016;Tellevik et al., 2016), and exotic bred in domestic animals (Seni et al., 2016). Data from this review found a significant number of ESBL-PE resisting non-beta-lactam antibiotics. More than 60% of the ESBL-PE isolated from humans, animals and the environment were resistant to trimethoprim-sulfamethoxazole and tetracycline/doxycycline, agents which are widely available and less costly in Tanzania and therefore more likely to be abused. In this review, the resistance to ciprofloxacin and gentamicin ranged from 46% to 75%. Despite of the established evidence that ESBL isolates tend to coresist, other classes of antimicrobials such as aminoglycosides (Doi & Arakawa, 2007), abuse of sulphur and tetracyclines in animals and human medicine (Sarmah, Meyer, & Boxall, 2006), and their longer persistence in the environment when disposed inappropriately (Kay, Blackwell, & Boxall, 2004;Monteiro & Boxall, 2010) might have allowed the selection of resistant strains and subsequent spread in the three interfaces. Similar to this review, a recent review from Kenya and another study from Uganda showed that there is no resistance to meropenem/imipenem (Kiiru et al., 2012;Najjuka et al., 2016). Moreover, studies in Kenya and Uganda also showed that piperacillin-tazobactam was effective in 60%-73% of isolates connoting its potential therapeutic alternative in the management of ESBL attributable infections (Kiiru et al., 2012;Najjuka et al., 2016).
In this review E. coli and K. pneumoniae were predominant ESBL-PE; an observation previously reported within and outside the African continent (Lautenbach, Patel, Bilker, Edelstein, & Fishman, 2001;Sonda et al., 2016;Storberg, 2014). The worldwide pandemic clone E. coli Nicolas-Chanoine et al., 2008, 2014 predominated in this review Mshana et al., 2016;Seni et al., 2016). Similar to the previous study carried out in three European countries to establish the clonality of ESBL-E. coli in humans, animals and food animals (Wu et al., 2013), bla CTX-M-group 1 in particular bla CTX-M-15 was the predominant allele in this review. Predominance of bla CTX-M-15 allele is in agreement with bacterial strains from drinking water sources in Tanzania, and clinical isolates in the North, West and South African countries such as Tunisia, Nigeria and South Africa (Lyimo et al., 2016;Storberg, 2014 T A B L E 2 The proportion of extendedspectrum beta-lactamase bacterial isolates expressing bla CTX-M-15 in humans, animals and the environment in Tanzania F I G U R E 5 Multiple Escherichia coli clones circulating between humans, animals and the environment in Tanzania. This figure was based on four articles which detected extended-spectrum betalactamase-producing Enterobacteriaceae sequence types Mshana et al., 2016;Seni et al., 2016) urine specimens (Sato et al., 2009;Storberg, 2014 (Mshana, Gerwing, et al., 2011).
As opposed to E. coli ST131 in humans, E. coli STs 617, 1303 and 2852 in animals and E. coli ST2852 in the environment predominated, connoting genotypic diversity of the strains across the three interfaces.
Surprisingly, the global ST131 previously reported (Mathers, Peirano, & Pitout, 2015;Nicolas-Chanoine et al., 2014;Price et al., 2013), as well as ST38 and ST2852 were found intersecting in the three interfaces calling for multisectoral genomic-based strategic intervention using One Health approach to combat the situation. Preponderance of occurrence of E. coli ST131 in humans as opposed to the animals and environmental sources was reiterated in this review, and also cautioned in another study in Tanzania (Madoshi et al., 2016). Despite the fact that none of the ESBL isolates were resistant to meropenem/imipenem in the present review, existence of the E. coli ST131 clone underscores a need for continuous AMR surveillance for potential occurrence of strains expressing carbepenemases in the future (Peirano et al., 2014).
Moreover, a previous study in Tanzania showed existence of carbapenemase genes among Gram-negative bacteria, reiterating further the need to monitor unprecedented phenotypic expression of these genes in the future (Mushi, Mshana, Imirzalioglu, & Bwanga, 2014).
The current review found more than 75% of the expressed ESBL genes in humans, animals and environmental isolates were located in conjugative IncF and IncY plasmids, with IncF plasmids cutting across all three interfaces. Similar observations have been observed in E. coli isolates from humans, animals and environmental sources in Europe, USA and other countries around the globe (Carattoli, 2009;Johnson et al., 2016;Wu et al., 2013). Although this review could not establish the transmission routes shared in the three interfaces, the similarity of the conjugative plasmids observed hint to the possibility of strong epidemiological link of the typed ESBL isolates, which worthy further scrutiny in the follow-up studies.

| CONCLUSIONS
Approximately 23% of humans, animals and environmental samples harbour ESBL-PE strains. The evidence of shared ESBL bla CTX-M-15 allele occurring in conjugative IncF plasmids, as well as ST38, ST131 and ST2852 clones in the three interfaces calls for One Health genomicdriven approaches to identify the resistome flow and subsequently prevent negative impacts attributable to ESBL-PE infections. In the light of these findings, ESBL transmission dynamics in the three interfaces and economic impacts should be potential areas for future research.