Fecal carriage of ESBL-producing E. coli and genetic characterization in rural children and livestock in the Somali region, Ethiopia: a one health approach

Description of study population

Out of the 366 children, only 346 children completed the questionnaire and were included in this study for further analysis. Pastoralists and agro-pastoralists were evenly represented in terms of age and sex among the enrolled children. Over half of the children showed a normal growth (61%), while a quarter (25%) of them were wasted, and 7.8% were stunted. Further, 5.2% of the children suffered concomitantly from stunting and wasting. Characteristics of the study group are summarized in Table 1.

Table 1 Characteristic of pastoralist and agro-pastoralist in the Adadle district, Somali region, Ethiopia

As summarized in the suppmentary Table S1, the two main sources of water for pastoralists and agro-pastoralists were rainwater/birkad (89%) and river water (95%), respectively. Most of pastoralist and agro-pastoralist (96%) used open space for defecation. The 14 households (4%) that had a toilet shared it with 28 households. Dumping waste in the street or open spaces within the compound was the predominant method of waste disposal in both pastoralist and agro-pastoralist societies (84%), while 16% opted to burn waste. Most mothers reported that they used water to wash the children’s hands (93%), while a small subfraction of mothers reported (4%) using water and soap. At the time of the sampling, more than half of the agro-pastoralists and pastoralists had soap, while 29% did not have soap very often and 13% never had soap in their house.

Nearly half of agro-pastoralist households (43%) and 12% of pastoralist households treated the water prior to consumption (mainly chlorination (98.9%)). The majority of agro-pastoralists possessed cattle (94.7%), donkeys (81.1%), goats (62.7%), sheep (53.3%), camels (10%) and chickens (10%). Among pastoralists, predominant livestock ownership included goats (87.6%), donkeys (58.8%), sheep (44.1%), camels (36.2%), and cattle (29.9%).

Phenotypic test results for ESBL E. coli carriage

A total of 609 fecal samples, comprising 366 from humans and 243 from animals (including 77 goats, 136 cows, and 30 camels) were analysed. For human isolates, 159 (43%) of the E. coli isolates were ESBL-producers (24.5% in pastoralists and 18.8% in agropastoralists). Furthermore, 7.8% of the E. coli isolates from goats (6/77) and 2.2% from cows (3/136) were ESBL-producers. Regarding animal ESBL-producing isolates, 7 (77.8%) were identified among animals of agro-pastoralists, while 2 (22.2%) were found among those of pastoralists.

Susceptibility pattern for ESBL-producing E. coli

For the human isolates, ESBL-producing E. coli strains exhibited complete resistance to several commonly used antibiotics including amoxicillin, cefotaxime, cefuroxime, ceftriaxone, cefazolin, and cefpodoxime. Almost all ESBL-producing E. coli isolates (98.7%) were resistant to ampicillin and 51.6% were resistant to tetracycline.

Regarding co-trimoxazole, 57.9% of the isolates were classified as resistant, 2.5% as intermediate, and 37.7% as susceptible. Among ESBL-producing E. coli, 47.8%, 42.1%, and 27.7% were non-susceptible, and 32.1%, 54.7%, and 25.2% were intermediate to aztreonam, cefipime, and ceftazidime, respectively. In addition, 15.7% and 12.6% of the isolates were resistant to azithromycin and ciprofloxacin. Ampicillin-sulbactam (8.8%), gentamicin (6.9%), and amoxicillin-clavulanic acid (5.9%) had the lowest rates of resistance. Intermediate susceptibility was observed in 16.5% for amoxicillin-clavulanic acid, 15.1% for ampicillin-salbactum, and 2.5% for gentamicin. All carbapenems (imipenem and ertapenem), which are considered the last-resort antibiotics for infections caused by ESBL-producing E. coli, were found to be effective against the ESBL-producing E. coli isolates, except for 1.3% of the isolates that had intermediate susceptibility to ertapenem. Additionally, using whole genome sequencing (WGS), none known carbapenem or colistin resistance genes were found in the E. coli isolates.

Additonally, nine ESBL isolates obtained from animal fecal samples exhibited complete (100%) resistance to ampicillin, cefotaxime, and amoxicillin. Furthermore, an 88.9% resistance rate was observed for cefepime, followed by cefazolin (55.6%), ceftriaxone (44.4%), cefuroxime (44.4%), and aztreonam and ampicillin-sulbactam, which both exhibited an equal resistance rate of 33.3%. In ESBL-producing E. coli isolates from animals, ampicillin-clavulanic acid, ceftazidime, gentamicin, and ciprofloxacin all exhibit an equal resistance rate of 11.1%. The data is summarized in Fig. 1.

Fig. 1figure 1

Antimicrobial resistance pattern of ESBL-producing E. coli isolated from the feces of children aged 2–5 years (A, 159/366) and from livestock (B, 9/243) in the Adadle district, Somali region, Ethiopia. Antimicrobial agents tested include: AMX (Amoxicillin), AMC (Amoxicillin-clavulanic acid), AMP (Ampicillin), ATM (Aztreonam), AZT (Azithromycin), CAZ (Ceftazidme), CIP (Ciprofloxacin), CN (Gentamicin), CPD (Cefpodoxime), CRO (Ceftriaxone), CTX (Cefotaxime), CCT (Cefotetan), ETP (Ertapenem), FEP (Cefepime), FOX (Cefoxitin), IMP (Imipenem), KZ (Cefazolin), SAM (Ampicillin-sulbactam), SXT (Co-trimoxazole), PIR (Piperacillin), TET (Tetracycline)

All ESBL-producing E. coli strains exhibited multi-drug resistance (MDR), rendering them non-susceptible to three or more antibiotic drug classes (Figure S1 in Supplementary File 1).

Thus, ESBL-carriage is very prevalent, especially in children, in a community setting and is much lower in the feces of livestock animals.

Predictors of ESBL-producing E. coli carriage

In the multivariable analysis, education was significantly associated with ESBL-producing E. coli. Children whose mothers or household heads were illiterate had twice the odds of carrying ESBL-producing E. coli compared to children whose mothers were formally educated (aOR = 2.65, 95% CI = 1.27–5.48). The odds of children who were both stunted and wasted were three time higher to harbor ESBL-producing E. coli compared to children with normal growth (aOR = 3.14, 95%CI = 1.02–9.07). Moreover, pastoralist children had 2.65 times higher odds of being colonized with ESBL producing E. coli compared to agro-pastoralist children (aOR = 2.65, 95% CI = 1.30–5.41). Counterintuitively, children who drank water treated with chlorine showed a positive association with ESBL-producing E. coli (aOR = 2.09, 95% CI = 1.10–3.98). Additionally, the possession of chicken increased the odds of infection with ESBL-producing E. coli five times (aOR = 5.13, 95% CI = 1.66–15.68). The sex and age of the child were not found to be significantly associated with infection with ESBL-producing E. coli. Similarly, although owing soap or washing hands with water and soap showed a trend to decrease the risk of contracting ESBL-producing E. coli, this association was not statistically significant (Fig. 2).

Fig. 2figure 2

Risk factors (multivariable model) associated with fecal carriage of ESBL-producing E. coli among children living in the Adadle district, Somali region, Ethiopia

Genetic characterization of ESBL strains (conventional PCR)

During PCR screening, we found that the blaCTX-M-15 gene was the most prevalent resistance gene in both human (82.8%) and animal (100%) isolates. The prevalence of the blaCTX-M-15 resistance gene was similar in isolates from the feces of pastoralists and agro-pastoralists, both almost at 80% (Figure S2 of the Supplementary Materials 1). Comparing the nine isolates from animals alongside those from children residing in the same household, we observed that only two households within the pastoralist group demonstrated simultaneous presence of blaCTX-M-15 in both their children and livestock (Figure S3).

Multi locus sequence types (MLST), phylogenetic groups and plasmid MLST

The 48 human E. coli isolate subjected to WGS analysis, were chosen based on their phenotypic profile. A sequence type (ST) by MLST using the Achtman scheme could be assigned to 44 isolates (91.7%), with five isolates having a single nucleotide polymorphism (SNP) in one gene (2 in adk, 2 in fumC, 1 in mdh). The most common ST was ST-2353 with five E. coli isolates, followed by ST-10 and ST-48 with three E. coli isolates each and ST-38, ST-450 and ST-4750 with two E. coli isolates each. All other 27 E. coli isolates had singular ST. A total of nine E. coli isolates were assigned to clonal complex ST10 (3 ST-10, 3 ST-48, 1 ST-227, 1 ST-378 and 1 ST-617) and two isolates to clonal complex ST38 (2 ST-38).

The predominant beta-lactam resistance gene among the 48 E. coli isolates was blaCTX-M-15, identified in 72.9% (35/48) of the whole-genome sequenced isolates, followed by blaTEM-1B identified in 47.9% (23/48), ampC beta-lactamase in 14.6% (7/48), blaOXA-1 in 8.3% (4/48) and blaCTX-M-55 in 4.2% of the isolates. The beta-lactam resistance genes blaCTX-M-14 and blaTEM-35 were the least prevalent, each detected in a single isolate.

Four E. coli isolates (8.3%) did not carry any known beta-lactam resistance genes, 20 E. coli isolates (41.7%) carried one, 19 E. coli isolates (39.6%) carried two and five E. coli isolates (10.4%) carried three beta-lactam resistance genes. Seventy-five percent (15/20) of E. coli isolates with one beta-lactam resistance gene carried only blaCTX-M-15, while 68.4% of E. coli isolates with two beta-lactam resistance genes carried both blaCTX-M-15 and blaTEM-1B.

Quinolone resistance conferred by mutations in the gyrA gene was detected in 33.3% (16/48) of E. coli isolates. Serine at position 83 was mutated to either Leucine (S83L, 10/16), Alanine (S83A, 3/16) or Valine (S83V, 3/16). In addition to S83L, three E. coli isolates had gyrA mutation D87N. Four E. coli isolates (8.3%) had parC mutation S57T and four E. coli isolates (8.3%) had parC mutation S80I combined with parE mutation S458A. The plasmid-encoded qnrS1 and qnrS13 genes were detected in 39.6% (19/48) and 2.1% (1/48) of the sequenced isolates, respectively.

Aminoglycoside resistance genes were present in 60.4% (29/48) of E. coli isolates. Four different aminoglycoside (3″) (9) adenylyltransferase (aadA) genes were detected in 15 E. coli isolates (31.3%), with aadA1 accounting for 66.7% (10/15), aadA2 and aadA24 for 13.3% (2/15) each and aadA5 for 6.7% (1/15) of the aadA genes found. Eighteen E. coli isolates (39.5%) had the plasmid-encoded aph(3″)-Ib and aph(6)-Id aminoglycoside resistance genes, one had only aph(3″)-Ib and one only aph(6)-Id. The aminoglycoside resistance genes aac(3)-IId, aac(6′)-Ib-cr, ant(2″)-Ia and aph(3′)-Ia were detected in one E. coli isolate each.

Sulfonamide resistance genes were detected in 62.5% of E. coli isolates (30/48), with sul1 in 30.0% (9/30), sul2 in 63.3% (19/30) and both genes in 6.7% (2/30) of the isolates. Trimethroprim resistance genes dfrA were detected in 58.3% of E. coli isolates (28/48), most commonly dfrA14 (9), followed by dfrA1 (7), dfrA5 (3), dfrA17 (3), dfrA7 (2), dfrA15 (2), dfrA8 (1) and dfrA19 (1). Tetracycline resistance genes tetA, tetB and tetD were detected in 35.4% (17/48), 16.7% (8/48) and 2.1% (1/48) of E. coli isolates, respectively. Macrolide resistance gene mphA was detected in 18.8% (9/48) of the isolates, with one isolate simultaneously carrying ermB. The results are summarized in figure S4.

Virulence genes and plasmids

A total of five different virulent genes were found in 15/48 of E. coli isolates, but none of the isolates were found to harbor Shigo-toxin genes in their genomes. The results revealed the presence of Enteroaggregative E. coli genes such as aggR (5/48), aggA (2/48) and aaiC (2/48). Additionally, both the heat-labile (elt) and heat-stable enterotoxigenic E. coli genes (est) were found in 7/48 and 8/48 E. coli isolates (Table S2).

The detection rate of IncFIB plasmid exhibited the highest frequency, found in 28/48 isolates, followed by IncFII (26/48), IncY (10/48), IncFIA (7/48) and Incl1-l (6/48). Combinations of multiple plasmids were observed, with IncFIB and IncFII being the most prevalent combination, present in 17/48 isolates. Specifically, 12 isolates carrying blaCTX-M-15 resistance genes encoded both IncFIB and IncFII plasmids. Three isolates carried a combination of three plasmids (IncFIB, IncFII, IncFIA), and single isolates carried four different plasmids (IncFIB, IncFII, IncFIA, IncY) (Fig. 3).

Fig. 3figure 3

Plasmid replicon profiles for 48 ESBL-producing E. coli isolated from 2 to 5 year old children in the Adadle district, Somali Region, Ethiopia. The presence of plasmid is shown in dark gray and absence in light gray

Comparison of genotypes and phenotypes

Resistance to different antibiotics was inferred from WGS data. The presence of specific genes (Fig. 4, in blue) indicated resistance to specific antibiotics (Fig. 4, in red). For example, the presence of at least one of the following genes, CTX-M-type, OXA-type, TEM-type, or ampC, indicated resistance to amoxicillin, ampicillin, aztreonam, cefotaxime, ceftriaxone, cefazolin, cefuroxime, and cefepodoxime. The same procedure was used for all other antibiotics, as indicated in Fig. 4.

Fig. 4figure 4

Concordance and discordance between genotype and phenotype for selected ESBL-producing E. coli isolates from Adadle district, Somali region, Ethiopia

The resistance results inferred from WGS data were then compared with the results of the phenotypic assay. We found high concordance rate (> 90%) between genotype and phenotype for azithromycin (93%), amoxicillin (91%), and ampicillin (91%). Conversely, chloramphenicol (87%), as well as several cephalosporins including cefotaxime, ceftriaxone, cefazolin, cefuroxime, and cefepodoxime, each showed a low concordance rate of 84%. Tetracycline (81%) and trimethoprim/sulfamethoxazole (72%) showed even lower concordance rates, with ciprofloxacin (63%) registering the lowest concordance rate among them. These discordances should be analysed in more detail by assessing more antibiotics and by mining for eventual new resistance mechanisms that could be encoded in these bacterial strains. The results are summarized in Fig. 4.

留言 (0)

沒有登入
gif