Extended-spectrum Beta-lactamase and AmpC beta-lactamases producing gram negative bacilli isolated from clinical specimens at International Clinical Laboratories, Addis Ababa, Ethiopia

Background Extended spectrum Beta-lactamases (ESBLs) and AmpC beta-lactamases (AmpC) are the common enzymes produced by gram negative bacilli, which are their main mechanisms of resistance to all generations of cephalosporins. Hence, this study aimed to determine the magnitude of ESBLs and AmpC producing gram negative bacilli (GNB) isolated from clinical specimens at International clinical Laboratories in Addis Ababa, Ethiopia. Methods A cross sectional study was conducted from January to May 2018. From different clinical specimens, 338 GNB were isolated and characterized. Bacterial species identification, antimicrobial susceptibility testing and screening for ESBLs and AmpC production were performed using Phoenix automated system (BD phoenix100). ESBLs production was confirmed using a combination disc method. All Cefoxitin resistant and confirmed ESBLs producing GNB were confirmed for AmpC beta-lactamases production by AmpC confirmatory Neo-Sensitabs discs (ROSCO tablet). Data were analyzed using SPSS version 20 software. Results E. coli 66.0% (224/338) followed by K. pneumoniae 12.1% (41/338) were GNB most frequently isolated. The overall magnitude of ESBLs producing GNB was 38.8% (131/338) and the extent of AmpC beta-lactamase producing GNB was 2.4% (8/338). Majority of ESBLs and AmpC beta-lactamases producing GNB were isolated from urine specimens 47.5% (116/338). Ampicillin (75.4%), amoxicillin with clavulanic acid (64.0%) and sulfamethoxazole-trimethoprim (55.6%) were most the antibiotics to which resistance was most commonly found. The multidrug resistance (MDR) level of GNB was 74.0% (250/338). Of ESBLs and AmpC beta-lactamases producing GNB, 99.3% were MDR (p < 0.05). Conclusion The high magnitude of ESBLs and AmpC beta-lactamases producing GNB calls the needs of strong intervention to minimize further occurrence and spread of such GNB. More importantly, the MDR level was high which suggests continuous monitoring & reviewing of antimicrobial policy in hospitals and the country at large.

In general, GNB that produce ESBLs and AmpC enzymes cause treatment difficulties and treatment failure, including morbidity and mortality. Therefore, detecting and determining the magnitude of ESBLs and AmpC is crucial for effective treatment and for prevention and control of these resistant bacteria. Additionally, adequate knowledge about the magnitude of ESBLs and AmpC is important to limit the spread and further prevalence of ESBLs and AmpC production in GNB. Hence, the present study aimed to determine the magnitude of ESBLs and AmpC producing GNB isolated from clinical specimens at International Clinical Laboratories in Addis Ababa, Ethiopia.

Methods
A cross sectional study was conducted from January 2018 to May 2018 at International Clinical Laboratories (ICL) in Addis Ababa, Ethiopia. ICL was established in 2004 and is one of the main clinical laboratories in Ethiopia which gives numerous services including microbiological analysis. Its microbiology department receives clinical specimens for microbiological analysis from different governmental and private hospitals in Addis Ababa. It is equipped with several automated machines including the Phoenix automated system that is used for bacterial identification and antimicrobial susceptibility testing. During the study period, a total of 338 GNB isolates were isolated from different clinical samples and included in this study for ESBLs and AmpC characterization and antimicrobial susceptibility testing.
The sample size was calculated based on single-population proportion using a previous study done in Ethiopia [16]. The socio-demographic data of patients was documented using the data collection sheet from the request form. The GNB isolates, the types of beta-lactamases and the antibiotics susceptibility pattern of the isolate were also recorded using a separate data collection sheet.

Isolation and identification of gram negative bacilli
The GNB isolates were obtained from various clinical specimens such as urine, pus, body fluids, sputum, stool, ear and eye discharges. The specimens were inoculated in to 5% sheep blood agar, XLD agar and MacConkey agar plates (all media were Oxoid Ltd, UK) and incubated at 37˚C for 18-24 hours. Each bacterium was characterized by colony appearance, size, consistency, shape, ability to ferment lactose and gram stain. All bacterial species were characterized and identified to species level using the Phoenix system (BD Diagnostic Systems, Oxford, UK) by adjusting the turbidity of bacterial suspension to 0.5 McFarland standards inoculum density. Bacteria suspension was prepared from pure colonies grown on primary isolation media and inoculated to the appropriate phoenix panel [17].

Screening of AmpC production
Guided by EUCAST, isolates showing reduced susceptibility to cefoxitin (MIC >8 μg/ml) were identified as a potential AmpC beta-lactamase producer. In addition, all ESBLs suspected isolates were screened for confirmation [19].

Confirmation of AmpC production
All cefoxitin non-susceptible isolates and ESBLs suspected isolates were checked for the presence of AmpC beta-lactamase using disc diffusion tablets Neo-Sensitabs (ROSCO, Taastrupgaardsvej 30, DK-2630 Taastrup, Denmark) one tablet with cefotaxime, one with ceftazidime and two tablets of cephalosporins combined with cloxacillin (AmpC inhibitor). An increase in the inhibition zone diameter of �5 mm for a combination disc versus ceftazidime or cefotaxime disk alone was confirmed as AmpC beta-lactamase producing GNB [20].

Data quality assurance
After preparation, culture media was checked for sterility by incubating 5% of them overnight and observing them for the presence of any bacterial growth. Quality control for the Phoenix machine and prepared media was performed using standard ATCC control strains.
For ESBL testing, K. pneumoniae ATCC 700603 was used as a (positive control) and E. coli ATCC 25922 was used as a (negative control) strains. Enterobacter cloacae (ATCC BAA 1143) and E. coli (ATCC 25922) were used as positive and negative QC strains for AmpC beta-lactamase producing GNB respectively. Before applying the 20% glycerol TSB for storage, it was QC tested for growth of E. coli ATCC 25922 standard strains. Data was re-assessed for its completeness, adequate recoding on the worksheet and in the SPSS software after entry.

Data analysis and interpretation
Data was analyzed using SPSS version 20 software. Descriptive statistics were summarized in tables and graphs. Proportions and actual number of AmpC and ESBLs producing GNB isolates were used to describe frequency of categorical variables. Continuous variables were described by mean and standard deviation.

Ethical considerations
The study was conducted after obtaining ethical clearance from the research and ethical review committee of the department of Medical Laboratory Sciences, School of Allied Health Sciences, College of Health Sciences; Addis Ababa University (DRERC/335/18/MLS). Additionally, an official permission letter was obtained from the study site.

Socio-demographic characteristics
During the study period, a total of 338 GNB isolates were isolated from various clinical specimens sent to ICL by government hospitals (n = 151) and private hospitals (n = 187). The majority of the GNB were isolated from urine 72.2% (244/338) followed by pus 18.6% (63/338) ( Table 1).
Among the total isolates, 58.3% (n = 197/338) were isolated from female patients and 41.7% (n = 141/338) were collected from male patients. E. coli (68.5%) and K. pneumoniae (14.2%) were the most frequently isolated bacteria from females. The majority of isolates 86 (25.4%) were obtained from patients above 61 years of age. The mean age of patients was 43.9 years and standard deviation 21.8 years ( Table 1).

Multi-drug resistance pattern of gram negative bacilli
Using the commonly used definition of multidrug resistance (MDR) as an organism being resistant to three or more classes of antibiotics, 74.0% (

Magnitude of ESBLs producing gram negative bacilli
From the total 338 isolated GNB, 135 were found to be suspicious for ESBLs production based on their MIC value of �2μg/ml for ceftazidime and/or cefotaxime as obtained from Phoenix system. Of these 135 GNB, 38.8% (131/338) of them were finally confirmed as ESBLs producers using the confirmatory combination disk method. A higher percentage of ESBL producing GNB was identified in K. pneumoniae 56.1% (n = 23/41) followed by E. coli (44.6%, 100/224), E. cloacae (36.4%, 4/11) and Citrobacter spp. (10.0%, 1/10) (Fig 1). The majority of these ESBLs producing GNB were detected in urine specimens and the proportion of ESBLs was significantly high among isolates from adult patients above 61 years of age (p < 0.05).
The MDR level of each Gram negative bacillus in relation to their ESBLs and AmpC production is summarized below (Fig 1). K.oxytoca (

Distribution of ESBLs and AmpC beta-lactamase producing gram negative bacilli with their MDR level among different specimens
From the specimens analyzed, ESBLs and/or AmpC producing GNB were found predominantly in urine 44.7% (109/244). The highest number of MDR isolates were found in wound specimens (81.0%, n = 51/249) ( Table 4). Of ESBLs and/or AmpC producing GNB, 97.8% (n = 137/140) were MDR and only 1.4% (n = 2/140) was non-beta-lactamases producing MDR Gram negative bacilli due to other resistance mechanisms. Being beta-lactamases producer has statistically significant association with the occurrence of MDR (p = 0.001).

Association of independent variables with magnitude of ESBLs and AmpC producing gram negative bacilli
Analysis of data using logistic regression model showed that the magnitude of ESBLs or AmpC producing GNB had statistically significant association with age group and specimen type. However, there was no statistical significance between sex and health facilities for acquisition of beta-

Antibiotics resistance pattern of gram negative bacilli
In the present study, the level of antimicrobial resistance of gram negative bacilli ranged from 0-75.4%. A higher resistance level was seen to ampicillin (75.4%), amoxicillin with clavulanic acid (64.0%), sulfamethoxazole-trimethoprim (55.6%), aztreonam and cefuroxime (48.8%), cefotaxime (47.0%), cefepime (45.6%), ceftriaxone (44.9%) and ceftazidime (44.1%). Comparable results were reported from studies performed in Gondar [16] and Debre Markos [21] in Ethiopia. Compared to findings done in Tanzania [22], Iran [23] and Sierra Leon [24] that reported a resistance level of >90% for ampicillin, amoxicillin with clavulanic acid, sulfamethoxazole-trimethoprim and ceftazidime, the resistance level recorded for these antibiotics in our study was a lowered finding that was <76%. Possible reasons for this difference can be the methods employed for antimicrobial testing, indiscriminate usage of antibiotics, patients condition and the nature of bacteria in these countries that were beta-lactamases producing bacteria in most cases.

Multi drug resistance pattern of gram negative bacilli
In the present study, the overall magnitude of MDR among all GNB isolates was 73.7%. There were similar results found in studies conducted in Gondar (68.0%) [25], Dessie (74.6%) [26] and Debre Markos: (72.2%) [21] in Ethiopia and also in Nepal (64.0%) [27]. Our study found a lowered MDR level than studies done in Gondar (87.4%) [16] and Bahir Dar (93.1%) [28] in Ethiopia and studies performed in Sierra Leone (85.7%) [24] and Nepal (96.8%) [29]. However, the magnitude of MDR we found among the GNB was higher than in previous studies done in Ethiopia (Jimma: 59.3%) [30] and Nepal (54.2%) [27]. The MDR level recorded in this study was considered as alarming because only a few treatment options remain for infections. Therefore, implementing strong infection control strategies is required to reduce the MDR burden.
However, the magnitude of ESBLs production among GNB in our study was lower than in other studies done in Addis Ababa in Ethiopia (57.7%) [36], Bahir Dar in Ethiopia (57.6%) [28] and north West Nigeria (58.0%) [37], southwest Uganda (89%) [38] and southeast Iran (53.8%) [23]. This wide variation might be due to differences in study population, type of specimen, sample size and the extent of antibiotic use.
In the current study, the predominant ESBLs producing isolates were K. pneumoniae (56.1%) and E. coli (43.8%). This finding was supported by previous studies done in Ethiopia in Addis Ababa: K. pneumoniae in 78.6% and E. coli in 52.2% [36] and in Bahir Dar: K. pneumoniae in 69.8% and E. coli in 58.2% [28]. Researches in other African countries also isolated these bacteria as predominant ESBLs. The previously mentioned study in north west Nigeria identified: K. pneumoniae in 62.9% and E. coli in 54.2% [37], the study in southwest Uganda: K. pneumoniae in 52% and E. coli in 44% [38], a study in Nairobi, Kenya: K. pneumoniae in 78.8% and E. coli in 60.7% [42], a study done in Uganda: K. pneumoniae in 72.7% and E. coli in 58.1% [43]. On the other side, our finding was in contrast with studies conducted in Sri Lanka: E. coli in 86.8% and K. pneumoniae in 13.1% [44] and India: E. coli in 50.14% and K. pneumoniae in 48.3% [45]. In these studies, E. coli was the predominant ESBLs producer instead of K. pneumoniae.

Magnitude of AmpC producing gram negative bacilli
AmpC beta-lactamase producing GNB have been responsible for several nosocomial outbreaks and high rate of treatment failure [46].
In present study, K. pneumoniae (7.3%, 3/41) and E. coli (2.2%, 5/224) were the principal AmpC producing GNB. This was also found in a study conducted in Turkey [50] and Spain [33]. This might be related to the fact that plasmid mediated AmpC beta-lactamases are seen in Enterobacteriaceae and these genes are easily horizontally transferred [13].
The current study also demonstrated co-existence of ESBLs and AmpC enzymes in five isolates 3.6% (5/139). This finding was comparable with studied conducted in Nigeria (6.04%) [40], South India (4.4%) [2] and India (9.9%) [9]. Simultaneous production of ESBLs and AmpC enzymes in a bacterium causes false-negative confirmatory tests for ESBLs production, because existence of plasmid mediated AmpC beta-lactamase enzymes can mask the presence of ESBLs [51]. Therefore, simultaneous detection of these enzymes is important to prevent the chance of missing an ESBLs.
Most of the AmpC producing GNB were resistant to the commonly used antibiotics which are known from several studies [40,51,54]. In our study, among cefoxitin resistance isolates, 10% (8/80) were AmpC producers which was in agreement with findings from studies done in Iran (5.1%) [47] and Turkey (8.7%) [50] but lower than findings from a study done in India (37.0%) [53]. The possible reasons for cefoxitin resistance in the absence of AmpC production might be due to other resistance mechanism like loose of permeability of porins [55].

Distribution of ESBLs and AmpC producing gram negative bacilli in different specimens
The present study shows that ESBLs and/or AmpC producing GNB were predominantly found in urine (44.7%, 109/244) and in pus (34.9%, 22/63). This might be due to the larger number of urine and pus samples included in this study. In this study most ESBLs producers were found in urine. Similar findings were reported in several countries: northwest of Nigeria (63.5%) [37], Uganda 64.9% [43], Ghana 66.7% [52], Sierra Leone 64.3% [24] and India (52.3% and 35%) [45,53]. Another study done in Adama, Ethiopia found pus as the major source of ESBLs producing GNB (53.0%) [39]. This difference is most likely related to the difference in the proportion of clinical specimens included, nature of study participants difference and risk factors.
All AmpC producing pathogens in our study were isolated from urine. Studies conducted in Turkey and Nigeria [50,57] had equivalent findings. Isolates from sputum (50.0%) were the predominant producers of AmpC beta-lactamase in a study done in Nigeria by Ogefere et al [46]. This indicated that the prevalence of AmpC beta-lactamase may differs significantly among bacteria obtained from different clinical specimens [34].

Conclusion and recommendation
In this study, a higher magnitude of ESBLs and AmpC beta-lactamases producing GNB was found. Majority of these beta-lactamase producing isolates were isolated from urine specimen. K. pneumoniae and E. coli were the most frequent identified ESBLs and AmpC beta-lactamase producing gram negative bacilli. These ESBLs and AmpC beta-lactamases producing isolates showed a high MDR level and were resistant for most available antibiotics. The most effective antibiotics for treatment of the identified gram negative ESBLs and AmpC beta-lactamases producers were amikacin, meropenem and imipenem. The occurrence of ESBLs and AmpC beta-lactamases requires strengthening of the antimicrobial resistance surveillance system and an effective antibiotic policy involving antibiotic restriction, combination therapy and infection control programs combined with good medical practices. Additionally, large scale studies that can assess the magnitude of ESBLs and AmpC beta-lactamases producers from a wider geographical perspective with more representative samples need to be done in the country.

Limitation of the study
• The magnitude of ESBLs and AmpC producing GNB from blood culture was not addressed because of service interruption due to down time of the Bactech 9050 machine during the study period.
Supporting information S1 Data. Raw data of the study. (PDF) S2 Data. Survey of extended-spectrum beta-lactamase and AmpC beta-lactamases producing gram negative bacilli isolated from clinical specimens at International Clinical Laboratories, Addis Ababa, Ethiopia. (DOCX)