Species, antibiotic susceptibility profiles and van gene frequencies among enterococci isolated from patients at Mulago National Referral Hospital in Kampala, Uganda

Background The increase in drug resistance to affordable antibiotics used to treat Gram positive bacterial infections has complicated the management of enterococcal infections. Resistance to vancomycin, one of the most powerful antibiotics, is of utmost concern as both intrinsic and acquired forms of resistance do occur in enterococci. This cross-sectional study aimed to determine the species, antibiotic susceptibility profiles and vanA/vanB gene frequencies among enterococci isolated from patients at Mulago Hospital in Kampala, Uganda. Methods Between November 2011 and October 2012, stool, urine, sputum and blood samples, as well as vaginal, endocervical, pus, ear and urethra swabs from 3229 patients were processed for isolation of bacteria, yielding 162 enterococci of which 115 were available for analysis (one isolate per specimen/patient). Species-level confirmation and susceptibility testing were determined with the Phoenix™ AST/ID Automated System, while vanA/vanB gene carriage was determined by PCR. Results Species-level identification revealed 72 isolates of E. faecalis, 20 E. gallinarum/casseliflavus, 5 E. faecium, 4 E. raffinosus and 2 isolates each for E. hirae and E. durans. Ten isolates could not be identified to species level. Antibiotic resistance was generally low especially to ampicillin, quinolones, nitrofurantoin, glycopeptides and linezolid, but high for erythromycin and tetracycline. Equally, vanA and vanB gene frequencies were low (i.e. 15.8 and 7.9%, respectively) and detected only in E. casseliflavus/gallinarum species that are intrinsically resistant to vancomycin. Vancomycin resistant isolates of E. faecalis and E. faecium were not detected. Conclusions Enterococcus species are frequent in clinical specimens at Mulago Hospital but they are highly susceptible to common antibiotics especially ampicillin. While vancomycin resistant enterococcal (VRE) isolates of E. faecium and E. faecalis are rare in the hospital and frequency of multidrug resistance is low, non-faecium and non-faecalis VRE isolates (i.e. E. gallinarum/casseliflavus) are frequent, some with VanA/VanB (high-level) vancomycin resistance. Therefore, species-level identification of enterococci is necessary in resource limited settings to guide infection control and treatment of enterococcal infections. Electronic supplementary material The online version of this article (10.1186/s12879-019-4136-7) contains supplementary material, which is available to authorized users.


Background
The enterococci are a diverse and versatile group of Gram positive, lactic acid producing bacteria, of which Enterococcus faecalis and E. faecium are the most predominant species. While they primarily inhabit the gastrointestinal tract of humans and other mammals, enterococci are also widely distributed in nature: they can be found in soil, water, dairy products and other foodstuffs, and on plants [1,2]. They are able to thrive under a variety of conditions and this creates a constant source of infection. Because of their inherent and acquired antibiotic resistance mechanisms, and ability to survive in harsh environments in community and hospital settings [3], enterococci have emerged as major causes of difficult-to-treat nosocomial and community acquired infections particularly biliary and urinary tract infections, bacteremia, endocarditis, intra-abdominal infections, pelvic infections, etc.
As mentioned, enterococci are intrinsically resistant to a variety of antibiotics especially cephalosporins, sulfonamides, oxacillin, ertapenem and perfloxacin. Specifically, E. faecalis and E. faecium are intrinsically resistant to cephalosporins, aminoglycosides, clindamycin, trimethoprim-sulfamethoxazole and fusidic acid; yet again E. faecalis is intrinsically resistant to quinupritin. Moreover, through mutation and horizontal gene transfer (HGT) processes, enterococci are able to acquire additional resistance mechanisms to key antibiotics notably tetracycline, erythromycin, fluoroquinolones, rifampicin, chloramphenicol, nitrofurantoin, fusidic acid, glycopeptides (vancomycin & teicoplanin) and to high concentrations of aminoglycosides and β-lactams [4]. While E. gallinarum and E. casseliflavus are less pathogenic than E. faecalis and E. faecium and they are highly susceptible to ampicillin, they are intrinsically resistant to one of the most potent antibiotics, vancomycin [4][5][6]. Furthermore, the hospitalization of patients and progress in medical technology and treatment, as well as increase in antibiotic usage, have contributed to a surge in infections due to multidrug resistant (MDR) enterococci and vancomycin resistant enterococci (VRE). VRE are of utmost concern in that vancomycin is a powerful antibiotic used to treat Gram positive bacterial infections yet, both intrinsic and acquired forms of resistance do occur in enterococci. VRE have become an important cause of serious invasive infections globally to such an extent that clinical microbiology laboratories are encouraged to speciate enterococcal isolates from hospitals and screen them for vancomycin resistance [4,6].
Glycopeptide resistance in enterococci is of two main phenotypes [4,6]. The first and most common vancomycin resistance phenotype among VRE is high-level resistance that occurs through acquisition of vanA & vanB genes, usually in E. faecium and E. faecalis, the species comprising majority of enterococcal infections especially bloodstream/invasive VRE infections. Intrinsic resistance to vancomycin also occurs in enterococci, characteristically in E. gallinarum and E. casseliflavus. These species possess the VanC enzymes, which synthesize pentapeptide peptidoglycan precursors ending in D-alanyl-Dserine that have reduced affinity for vancomycin hence, they are the source of intrinsic vancomycin resistance in E. gallinarum/casseliflavus [4,6]. Note, high-level vancomycin resistance of the VanA/VanB phenotype is transferrable and it occurs in E. gallinarum/casseliflavus [4][5][6]. Overall, VRE other than E. faecium and E. faecalis are infrequently associated with infections but their frequencies are likely to be underreported in the developing countries.
Henceforth, it has become necessary to screen enterococci isolated from hospitalized patients for vancomycin resistance, especially isolates from patients with endocarditis and meningitis, and isolates from sterile body parts [4]. VRE speciation is particularly important as it allows differentiation of vanC carrying organisms and avoid misidentifying them as VanA/VanB VRE. Furthermore, as colonization of the gut and other body parts by the enterococci is a key predisposing factor for infection and it is the main source of endogenous infection by enterococci [2], characterizing enterococci from clinical specimens is proxy for understanding enterococcal colonization, and it generates insight into their infection dynamics in specific settings.
Although enterococci are amongst the leading causes of urinary tract infections (UTIs) and bacteremia worldwide [7], their significance in Uganda is not clear as there are no clearly defined rates for enterococcal infections in this country. All the same, preliminary estimates for enterococcal infections reveal high rates for UTIs and surgical site infections (SSIs) in Uganda [8,9]. The aim of this study was to define the species and antibiotic susceptibility profiles of enterococci isolated from patients at Mulago National Referral Hospital in Kampala, Uganda. We also estimated the vanA and vanB gene frequencies among the isolates.  Table S1. These specimens were processed on request by clinicians during routine clinical investigations.

Study setting, specimens and isolates
Of the 3229 specimens, Gram positive and catalase negative isolates grew from a total of 196 specimens (one isolate per specimen). All the 196 Gram positive and catalase negative bacterial isolates had a diplococcal chain appearance under microscopy, and they were subjected to growth on bile esculin agar (BEA), a selective differential medium used to identify members of the genus Enterococcus. Additionally, a 6.5% (w/v) sodium chloride tolerance test was carried out in order to differentiate enterococci from group D streptococci. Following these two procedures, a total of 162 pure isolates (one per specimen) were presumptively confirmed to be enterococci and stored at − 80°C.
Presumptive species-level identification was based on motility and pigmentation, and formation of acid in mannitol, sorbitol, sucrose, arabinose and pyruvate broths [6]. Briefly for the pyruvic acid test, pyruvate broth was inoculated with a single colony from a 24 h pure culture, incubated aerobically at 35°C with a loose cap and examined daily for 3-5 days. Acid production was confirmed by color change in the bromothymol indicator from blue-green to yellow, which was interpreted as positive for E. faecalis. E. faecalis ATCC51299 was used as the positive control for the pyruvate test. For the arabinose test, Brain Heart Infusion (BHI) broth containing arabinose and bromocresol purple indicator was inoculated with a single colony from a 24 h pure culture, and incubated overnight at 35°C. Color change in the media to yellow was interpreted as positive for E. faecium [10]. For motility test, microscopy was done to presumptively differentiate E. gallinarum from other enterococci.

Antibiotic susceptibility profiles and van gene frequencies
Antibiotic resistance was generally low for all the species especially to ampicillin, quinolones (ciprofloxacin & moxifloxacin), gentamicin-syn, nitrofurantoin, glycopeptides, linezolid and daptomycin but high for erythromycin and tetracycline, Table 2. All E. faecalis isolates were susceptible to nitrofurantoin, vancomycin, teicoplanin and linezolid but resistant to erythromycin and tetracycline, Table 2. Further, all but one isolate of E. faecalis that were intermediate or resistant to ciprofloxacin were also resistant to tetracycline, and all except two were resistant to erythromycin. Of the three isolates that were resistant to gentamicin-syn, two were E. faecalis (Table 2) one of which exhibited multiple resistance to several antibiotics (i.e. ampicillin, erythromycin, nitrofurantoin, ciprofloxacin, tetracycline) but susceptible to teicoplanin, vancomycin and daptomycin. For E. faecium, all the five isolates were susceptible to gentamicin-syn, ampicillin, vancomycin, teicoplanin, linezolid and moxifloxacin however, four were non-susceptible to daptomycin. One isolate of E. faecium was resistant to both erythromycin and nitrofurantoin and intermediate to ciprofloxacin.  Furthermore, all the other species combined (i.e. E. casseliflavus/gallinarum, E. raffinosus, E. hirae, E. durans and the 10 isolates that were not identified to species level) exhibited varied susceptibilities to antimicrobials with a total of 31 isolates being non-susceptible to daptomycin. Overall, vancomycin resistance in this study was detected only in the 20 isolates of E. casseliflavus/ gallinarum ( Table 2) that are intrinsically resistant to this drug. Likewise, resistance to teicoplanin was detected in only E. casseliflavus/gallinarum at 15% (3/20); however, the teicoplanin resistant isolates were susceptible to ampicillin, erythromycin, nitrofurantoin, tetracycline and gentamicin-syn. Once more, the only linezolid resistant isolate was E. casseliflavus/gallinarum; it was also resistant to vancomycin, teicoplanin and moxifloxacin but susceptible to ampicillin, tetracycline, nitrofurantoin and gentamicin-syn.
van gene frequencies and multidrug resistance phenotype Generally the vanA and vanB gene frequencies were low i.e. 15.8 and 7.9% respectively, and detected in only E. casseliflavus/gallinarum species, Table 3 & Additional file 2: Figure S1. The species and resistance patterns of enterococci resistant to two or more antibiotics are summarized in Table 4. Regarding resistance to antibiotics to which enterococci are commonly susceptible (i.e. erythromycin, tetracycline, ciprofloxacin, moxifloxacin, ampicillin, nitrofurantoin, vancomycin, teicoplanin, linezolid, daptomycin), only 11 isolates (9.6%, 11/115) were MDR and almost all the MDR patterns involved resistance to tetracycline, Table 4. Majority of MDR isolates belonged to E. faecalis (five isolates) and E. casseliflavus/gallinarum (four isolates) while the remaining two isolates were E. raffinosus and E. durans. Besides MDR, the most common resistance pattern among isolates resistant to two or more antibiotics involved resistance to erythromycin and tetracycline, Table 4. Lastly, cluster analysis of the Rep-PCR fingerprints revealed no evidence of clonal spread/transmission of enterococci in the Mulago Hospital setting (not shown).

Discussion
In this study, one of the most predominant Enterococcus species, E. faecalis [4,13,14], was the most frequent at 62.6%. However E. faecium, another key enterococcus species worldwide especially in hospital settings, was not frequent. The reported low frequency for E. faecium and several other enterococcus species in this study (E. durans, E. hirae and E. raffinosus) is in line with global reports for these species [4] in that up to 80-90% of the enterococcal infections in humans are attributed to E. faecalis, 10-15% E. faecium and < 5% combined other species [6]. Note, the proportion of E. faecium in this study was twofold lower than expected. While this could have resulted from a potential sampling bias, E. faecium being significantly less frequent than E. faecalis at Mulago Hospital has previously been reported i.e. prevalence of 8.7% (2/23) and 91.3% (21/23) for E. faecium and E. faecalis, respectively [9].
Interestingly, the second most frequent enterococcus species in this study were the E. gallinarum/casseliflavus  x -MICs for these species (see "others") are not reported by the Phoenix system MDR, multidrug resistant (isolate resistant to three or more antibiotic classes) "Others" denotes E. casseliflavus/gallinarum, E. raffinosus, E. hirae, E. durans, and the 10 isolates that were not identified to species level species at 17.4%, which is higher than expected from global reports for these species (i.e. ≤2%) [5,6]. Although E. gallinarum/casseliflavus are infrequently isolated from clinical specimens, especially in the developed countries, several reports have implicated them as causes of serious invasive infections [5,6]. These species colonize the gut of hospitalized and non-hospitalized people with overall colonization rates ranging from 5.7-12.1% [6]. The current study, and recent findings from Ethiopia [15] and Sri Lanka [16], suggest that the colonization rate for these species is high in developing countries. However, as we investigated isolates from symptomatic individuals, our findings should not be used to draw inferences on enterococcus colonization in the wider population. The clinical and epidemiological features of cases of E. casseliflavus/ gallinarum bacteremia in the United States were reported [5,6], with underlying conditions such as immunodeficiency, malignancy, receipt of transplant etc. reported in up to 95% of the patients [5,6]. Of all the risk factors investigated only the immunocompromised status was a significant predictor of mortality in patients with E. gallinarum/casseliflavus bacteremia [5]. Almost all the enterococci in this study were susceptible to ampicillin (only three isolates were resistant) and this is in agreement with the finding that enterococci are generally susceptible to ampicillin [4]. Note, ampicillin resistance in this study was significantly lower than rates from Ethiopia [15] and India [17], but this could be due to differences in methods for drug susceptibity testing e.g. we used MICs by an automated instrument that are generally more efficient compared to the disk diffusion methods used by investigators in Ethiopia [15]. As ampicillin is the drug of choice in treatment of enterococcal infections, the high susceptibility of isolates implies that this cheap antibiotic is effective in treatment of enterococcal infections in Uganda. Only three isolates exhibited high-level resistance to gentamicin, two of which were E. faecalis and these are few compared to rates  from Ethiopia [15]. At the moment, high-level gentamicin resistance in enterococci may not be a cause for concern in Uganda however, continuous surveillance in hospital settings is necessary to monitor this trend. Further, several isolates in this study were non-susceptible to daptomycin, and this could be attributed to tolerance ("the ability of the organism to survive levels of drugs well in excess of the MIC" [4]), a common finding among clinical isolates of enterococci [4].
All the E. faecalis and E. faecium isolates in this study were susceptible to vancomycin and the only VRE isolates detected were E. gallinarum/casseliflavus species that are inherently resistant to vancomycin. Elsewhere, investigators in Spain who studied the antibiotic resistance of enterococci from healthy breastfed infants also did not detect VRE isolates of E. faecalis and E. faceicum [18]. As the VRE isolates (i.e. E. gallinarum/casseliflavus) in this study were susceptible to ampicillin, this antibiotic should be the drug of choice for treating enterococcal infections in Uganda as increased use of vancomycin may lead to selection of glycopeptide resistant enterococci in a resource-limited setting. It is important to note that lack of VRE isolates of E. faecalis/ faecium in this study may not imply they are rare in Uganda: a VRE isolate of E. faecalis was recovered from bovine milk samples in Kampala [19] and from a patient with a SSI at Mulago Hospital [9].
The vanA and vanB genes were not detected in E. faecalis and E. faecium, which was expected as these genes encode high-level vancomycin resistance [20] that was not detected in E. faecalis/faecium isolates. Yet, vanA/ vanB occurred in E. gallinarum/casseliflavus; although vancomycin resistance in these species is reported to be of the low-level type [4][5][6], their detection implies that acquisition of the vanA/vanB genes occurred, most likely in food animals as vancomycin resistant E. faecalis and E. gallinarum/casseliflavus were recovered from bovine clinical specimens [19]. As high-level vancomycin resistance of the VanA and VanB phenotypes is transferrable, and it has been reported in E. gallinarum and E. casseliflavus which are part of the normal flora in mammals [4], individuals in Uganda could be at risk of being colonized by such isolates via transmission from animals [4,6]. Overall, our data reveals that speciation and   [4]. Moreover, because E. gallinarum/casseliflavus appear more frequent in human specimens [15,16] and bovine specimens [19] in lowincome countries compared to the developed countries [4][5][6], species-level identification of enterococci is key for instituting proper infection control measures and timely antibiotic therapy [6]. While the MDR frequency was low in this study, the isolates exhibited unique resistance patterns, for example ERY-GEN-AMP-CIP-MXF-TET (E. faecalis); ERY-AMP-NIT-MXF-TET (E. durans); VAN-TEC-LIZ-MXF (E. gallinarum/casseliflavus); VAN-TEC (E. gallinarum/casseliflavus). Thus, we have reported E. gallinarum/casseliflavus isolates that were resistant to both vancomycin and teicoplanin though they were susceptible to ampicillin. Antibiotic stewardship and strengthening of the infection control practices is necessary to prevent acquisition of ampicillin resistance in such strains. Lastly, cluster analysis of the Rep-PCR fingerprints revealed no evidence of clonal spread/transmission of enterococci in the Mulago Hospital setting though we used an inferior approach to infer genetic relatedness among isolates. However, this finding is in accordance with reports that VRE isolates in certain settings (e.g. the United States) are genetically diverse [4].
There were certain limitations in this study: First, a total of 47 stored enterococcal isolates mainly from vaginal swabs, urine and stool were not recovered on subculturing. However, as we succeeded in sub-culturing isolates from specimens where enterococci grew most (see Additional file 1: Table S1), we believe we avoided a potential selection bias. Second, we could not rule-out contamination: Given the high rate of colonization of health care workers by the enterococci, some of the isolates might have been not clinically relevant. Third, we did not investigate E. gallinarum/casseliflavus for vanC genes as these are chromosomally encoded hence inherent in these species. As well, our approach for isolate genetic relatedness was not robust and we recommend studies with superior approaches such as whole genome sequencing.

Conclusions
Enterococci are frequent in clinical specimens at Mulago National Referral Hospital but they are susceptible to antibiotics commonly used to treat their infections e.g. ampicillin. As well, the frequency of MDR enterococci in the hospital is low and VRE isolates of E. faecium and E. faecalis are rare. However, non-faecium and non-faecalis VRE (E. gallinarum/casseliflavus) are frequent some with VanA and VanB high-level resistance to vancomycin. Thus, species-level identification of VRE isolates is important for instituting effective infection control measures that will curb the spread of VRE in the hospital.

Additional files
Additional file 1: Table S1. Samples processed for isolation of bacteria.