Spectrum of antibiotic resistant bacteria and fungi isolated from chronically infected wounds in a rural district hospital in Ghana

Background Chronic infected wounds are generally difficult to manage and treatment can be particularly challenging in resource-limited settings where diagnostic testing is not readily available. In this study, the epidemiology of microbial pathogens in chronically infected wounds in rural Ghana was assessed to support therapeutic choices for physicians. Methods Culture-based bacterial diagnostics including antimicrobial resistance testing were performed on samples collected from patients with chronic wounds at a hospital in Asante Akim North Municipality, Ghana. Fungal detection was performed by broad-range fungal PCR and sequencing of amplicons. Results In total, 105 patients were enrolled in the study, from which 207 potential bacterial pathogens were isolated. Enterobacteriaceae (n = 84, 41%) constituted the most frequently isolated group of pathogens. On species level, Pseudomonas aeruginosa (n = 50, 24%) and Staphylococcus aureus (n = 28, 14%) were predominant. High resistance rates were documented, comprising 29% methicillin resistance in S. aureus as well as resistance to 3rd generation cephalosporins and fluoroquinolones in 33% and 58% of Enterobacteriaceae, respectively. One P. aeruginosa strain with carbapenem resistance was identified. The most frequently detected fungi were Candida tropicalis. Conclusions The pathogen distribution in chronic wounds in rural Ghana matched the internationally observed patterns with a predominance of P. aeruginosa and S. aureus. Very high resistance rates discourage antibiotic therapy but suggest an urgent need for microbiological diagnostic approaches, including antimicrobial resistance testing to guide the management of patients with chronic wounds in Ghana.

Besides bacteria, fungi may be part of polymicrobial communities on wounds. In addition, specific fungal pathogens may cause chronic skin infections, including mycetoma and chromoblastomycosis often manifesting as slow-healing wounds [3,4].
In resource-limited countries with inadequate diagnostic facilities, the spectrum of microorganisms causing such infections and their antibiotic resistance patterns are only superficially understood [5][6][7][8].
Pathogen-specific treatment for severe wounds is difficult if the disease-causing agent(s) remain(s) unknown. In addition, treatment success is under serious threat due to the rise of antimicrobial resistances [9]. Poor quality medicine and (over)-use of antibiotics without being able to consider the antimicrobial sensitivity patterns of causative pathogens does not only lead to treatment failures but also foster the development of antibiotic resistance further. These limitations hamper specific treatment, which especially applies to chronic wounds, for which standard treatment strategies are poorly defined and ineffective.
This study aims to investigate the microbial composition (both bacterial and fungal) of infected chronic wounds in adults presenting to the Agogo Presbyterian Hospital (APH) in rural Ghana.

Study site and study participants
The study was conducted at the Outpatient Department of the APH, in the Asante Akim North Municipality of Ghana. Patients aged ≥ 15 years with an infected wound (at least one of the signs and symptoms of infection: purulent discharge from wound/pain or tenderness, localised swelling or redness/heat, loss of function (functio laesa)), which has failed to proceed through a reparative process over a period of one month, with or without antibiotic treatment, were eligible for enrolment.
Excluded were patients with burn-, surgical-bite-or Buruli ulcer associated wounds in order to minimise associations of particular types of wounds with specific pathogens. 5 informed consent was obtained before enrolment.

Sample Collection
For sample collection two sterile swabs (ESwab™, COPAN Diagnostics, Murrieta, USA), one with and one without transport medium, were used. Necrotic material was removed with a sterile cotton swab (CLASSIQSwab™, COPAN Diagnostics). For sample collection, the wound was cleaned with sterile 0.9% saline. Following this, the active part of the wound below the necrotic tissue at the edge of the wound and the wound base was swabbed. For this purpose, the swab was rolled deep into the wound. When there was more than one wound at the same location, the largest wound was sampled.
Within six hours, the samples were transported in a cool box to the microbiology laboratory of the Kumasi Centre for Collaborative Research in Tropical Medicine, Ghana.

Bacterial detection and identification
The swab with transport medium was streaked on Columbia agar enriched with 5% sheep blood, Columbia CNA agar with 5% sheep blood and MacConkey III (Oxoid, Hampshire, United Kingdom).
Aerobic blood agar was incubated in CO 2 . MacConkey and CNA agar were incubated in normal atmosphere at 35-37 °C. All plates were incubated for 18-24 hours. CNA agar was incubated until positive or for a maximum period of five days. Bacterial strains were identified by colony morphology, Gram stain and standard biochemical methods and stored in microbanks at -80 °C until transportation on dry ice to the Bernhard Nocht Institute in Hamburg, Germany for further analysis.
Environmental bacteria and bacteria belonging to the skin flora (e.g. coagulase-negative staphylococci, Micrococcus spp., coryneform bacteria and Bacillus spp. other than Bacillus anthracis) were classified as contaminants. At the Institute for Medical Microbiology in Göttingen, Germany, species identification was confirmed using the MALDI Biotyper 3.0 (Bruker Daltonics, Bremen, Germany).

Antibiotic susceptibility testing
Susceptibility to locally-available antibiotics was tested by the disk diffusion method and interpreted following the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines v.6.0 (http://www.eucast.org). Quality control of susceptibility testing was performed according to EUCAST (QC table v.5). At the University Medical Centre, Göttingen, Germany antibiotic susceptibility testing was confirmed using the automated VITEK 2 system (bio-Mérieux, Marcy-l'Étoile, France). Quality control was performed with the following ATCC reference strains: Pseudomonas aeruginosa 27853, Staphylococcus aureus 29213 and Escherichia coli 25922.

Fungal detection and identification
DNA was extracted from the swab (without transport medium) within 12 hours according to manufacturer's guidelines using the BIOstic Bacteremia DNA Isolation kit (QIAGEN, Hilden, Germany).
DNA extracts were stored at -80 °C until polymerase chain reaction (PCR) amplification for clinically important fungi. Fungal DNA was amplified using a broad-range fungal PCR targeting the 28S rRNA gene using Eva Green with melt curve analysis in an ABI 7500 qPCR machine as described previously [10]. Samples were tested in duplicate. Positive fungal PCR was defined as amplification of DNA of identical melt curves in both duplicates with a threshold cycle below 45. Template controls (n = 8) were included in each run to document potential contamination with fungal DNA. Amplicons of PCR positive samples were sequenced by Sanger sequencing. Fungi were identified by BLAST search using Genbank. Identification of fungi required identity > 98% over the amplicon length.
In addition, a qPCR assay targeting the human 18S rRNA was performed to document successful sampling and DNA extraction. Furthermore, potential inhibition of the PCR reaction was screened by an internal amplification control, amplifying an artificial plasmid.

Statistical analyses
Categorical variables were described using frequencies and their proportion and continuous variables using the median and the interquartile range (IQR). Prevalence ratios (PRs) along with their 95%confidence intervals (CIs) were calculated to show associations between dichotomous variables.
Missing values were excluded from the analysis; hence, in some calculations the denominator differs.
All analyses were conducted using Stata Statistical Software 14 (StataCorp LLC, College Station, TX).

Study group
One hundred and five outpatients were enrolled between January 2016 and November 2016.
The detected potential pathogens are listed in Table S1. The majority of isolates were Gram-negative (n = 143; 69%) bacteria. Most were Enterobacteriaceae (n = 84; 41%) and Pseudomonas aeruginosa (n = 50; 24%) followed by Staphylococcus aureus (n = 28; 14%). In general, the bacterial composition was comparable between mono-and polymicrobial infections (Fig. 1). The most frequently detected pathogen in mono-and polymicrobial infections was Pseudomonas aeruginosa. In patients with one or two isolates detected, Staphylococcus aureus was the second most common pathogen found. Patient and wound characteristics stratified by the presence or absence of Gram-negative bacteria are summarised in Table 1, indicating that characteristics were comparable between both groups.

Antibiotic susceptibility testing
Antibiotic susceptibility results of the most common bacterial species using the VITEK 2 system are shown in Table 2

Discussion
In this study, we assessed the microbial composition of chronically infected wounds in rural Ghana.
The bacterial spectrum seen was in line with findings from previous studies conducted in another region of Ghana, in Ethiopia and in Tanzania [1,[11][12][13][14][15].
Altogether, Gram-negative bacteria dominated quantitatively. Enterobacteriaceae, Pseudomonas aeruginosa and Staphylococcus aureus were the most frequently isolated organisms. Previous assessments both from America and Europe as well as from Africa confirm Staphylococcus aureus and Pseudomonas aeruginosa as the leading pathogens isolated from chronic wounds. Both species express virulence factors and surface proteins negatively affecting wound healing [1,9,16,17].
Pathogenicity in chronic wounds is further increased by co-infections of Staphylococcus aureus and Pseudomonas aeruginosa [16].
Biofilms caused by Pseudomonas aeruginosa play important roles by maintaining chronic wound infections thus preventing healing. Such biofilms are also known to show high adherence to biological surfaces, a phenomenon which limits the sensitivity of swabbing-based sampling with subsequent culture-based diagnostic approaches in comparison to molecular biological approaches [18].
Consequently, it is likely that the true prevalence of Pseudomonas aeruginosa was even higher.
Although Pseudomonas aeruginosa followed by Staphylococcus aureus were the most common bacteria identified on species level, the most frequent isolates were Enterobacteriaceae. The estimation of their clinical relevance is much more difficult. Although it is generally accepted that Enterobacteriaceae like Escherichia coli and others might play a role in wound infections [17,19].
inhabitants of tropical or subtropical climate zones were shown to have high colonization rates of skin and mucous membranes with Gram-negative rod-shaped bacteria [20]. High frequencies were also reported from patients, students, and health-care workers in Madagascar [21]. This makes the discrimination between wound colonization and wound infection challenging. From the less frequently isolated bacteria, beta-hemolytic streptococci including Streptococcus pyogenes are highly likely to be etiologically relevant for the assessed chronic wound infections [22], while enterococci are frequent colonizers of uncertain clinical relevance in wounds [23]. In mono-and polymicrobial infections, Pseudomonas aeruginosa was the most common bacterial specie. Others, including Enterobacteriaceae and enterococci were rarely observed in cases in which only one microorganism was isolated, stressing the likelihood that they are probably contamination flora or minor components of polymicrobial infections.
The observed antimicrobial resistance patterns are of interest. The recorded 29% MRSA is in discordance with a Ghanaian study on wound infections from 2014, where no MRSA were identified from infected wounds [14]. This suggests a considerable increase of MRSA within less than a decade; although it has to be considered that selection processes in chronic wounds after repeated antimicrobial therapy are likely. High frequencies of 3rd generation cephalosporin and fluoroquinolone resistance but absence of carbapenem resistance in Enterobacteriaceae isolated from wounds in Ghana has, in contrast, been described previously [14]. The increase of fluoroquinolone resistance in Ghana has recently been described [24]. significantly reducing the value of this group of antibiotics for treatment of severe infections. In addition, we found one carbapenem-resistant Pseudomonas aeruginosa. Carbapenem-resistance is still comparatively rare in Ghana: 2.9% in Gram-negative rodshaped bacteria, predominantly in Pseudomonas aeruginosa and Acinetobacter baumannii [25]. In summary, the observed resistance patterns make any rational antimicrobial therapy challenging, indicating a need for routine diagnostics including antibiotic susceptibility testing. However, antibiotic resistance seen in bacterial isolates from chronic wounds is likely to overestimate the true resistance rate in clinical isolates, because selection processes due to repeated attempts of antibiotic treatment have to be considered as likely.
In regards to fungi isolated from chronic wounds, information on the immune status of the patients was not assessed. Hence potential etiological relevance remains speculative. In general, fungal infections have been seen to cause skin infection including mycetoma that may manifest as chronic wounds after trauma [26]. Detection of causative fungal pathogens requires prolonged culture on fungus-specific media or demonstration of fungal elements in tissues. The performance of broadrange fungal PCR to amplify fungal DNA for the diagnosis of these infections has not been evaluated [27]. Our results suggest that the applied approach may be more likely to amplify DNA of yeasts that may be normal skin flora or microbial communities of chronic wounds. Specific PCR assays may be more sensitive for the amplification of specific fungal pathogens such as Madurella mycetomatis [26].
Limitations of this study include geographic restrictions of sampling, reducing the representativeness of the study for the whole of Ghana. In addition, swabbing of neighboring sites of intact skin to better differentiate infection from colonization [28] was not performed. Also, more sensitive molecular diagnostic methods were not applied, making it likely that several components of mixed infections will have been missed [18]. Also, despite their potential involvement in chronic wound infections, no anaerobic bacterial culture was performed.

Conclusions
As in other studies elsewhere, Pseudomonas aeruginosa and Staphylococcus aureus were the most frequently detected isolates and thus the most important species to be considered by physicians for therapeutic decisions.
In addition, high rates of antibiotic resistance are to be expected for isolates from chronic wounds in Ghana. This stresses the need for routine bacterial diagnostics including antimicrobial resistance testing prior to targeted antimicrobial therapy if antibiotic treatment is required and locally disinfecting or surgical procedures are considered insufficient. If routine bacteriology is impossible for logistical or infrastructural reasons, the observed lack of carbapenem-resistance suggests that antibiotic substances, which are available for intravenous application only, may still have good chances of showing clinical effect. However, decision-makers should be aware that empiric treatment further increases antibiotic resistance and is thus a less sustainable approach.
Declarations Figure 1 Proportion of Enterobacteriaceae and Pseudomonas aeruginosa resistant to locally available antibiotics.

Figure 2
Pathogen distribution in mono-and polymicrobial bacterial wound infection

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.