Microbiological profile of diabetic foot ulcers in Kuwait

Introduction Diabetic foot ulcers (DFU) and infection (DFI) are a major diabetes-related problem around the world due to the high prevalence of diabetes in the population. The aim of our study was to determine the microbiological profile of infected ulcers in patients attending Dasman Diabetes Institute (DDI) clinics in Kuwait and to analyze the distribution of microbial isolates according to wound grade, sex, age and diabetes control. Methods We collected and analyzed clinical data and samples from 513 diabetic patients with foot ulcers referred to our podiatry clinic at DDI from Jan 2011 till Dec 2017. Results We show a higher prevalence of DFU in men than in women, and a greater percentage of DFU occurred in men at an earlier age (p<0.05). Only about half of the DFU were clinically infected (49.3%) but 92% of DFU showed bacterial growth in the microbiological lab analysis. In addition, we isolated more monomicrobial (57.3%) than polymicrobial (34.8%) DFI and representing an average of 1.30 pathogens per patient. The presence of Gram-positive and Gram-negative strains was comparable between men and women regardless their age or glucose levels. Interestingly, more Gram-positive strains are present in ulcers without ischemia while more Gram-negative strains are present in ulcers with ischemia (p<0.05). While Staphylococcus aureus was common in infected ulcers without ischemia, Pseudomonas aeruginosa was predominant in ulcers with infection and ischemia, regardless of ulcer depth. Finally, a higher percentage of women has controlled HbA1c levels (19.41% versus 11.95% in men) and more women in this group displayed non-infected wounds (60.6% and 43.90% for women and men, respectively). Conclusion Our results provide an updated picture of the DFI patterns and antibiotics resistance in patients attending Dasman Diabetes Institute (DDI) clinics in Kuwait which might help in adopting the appropriate treatment of infected foot and improving clinical outcomes.


Results
We show a higher prevalence of DFU in men than in women, and a greater percentage of DFU occurred in men at an earlier age (p<0.05). Only about half of the DFU were clinically infected (49.3%) but 92% of DFU showed bacterial growth in the microbiological lab analysis. In addition, we isolated more monomicrobial (57.3%) than polymicrobial (34.8%) DFI and representing an average of 1.30 pathogens per patient. The presence of Gram-positive and Gram-negative strains was comparable between men and women regardless their age or glucose levels. Interestingly, more Gram-positive strains are present in ulcers without ischemia while more Gramnegative strains are present in ulcers with ischemia (p<0.05). While Staphylococcus aureus was common in infected ulcers without ischemia, Pseudomonas aeruginosa was predominant in ulcers with infection and ischemia, regardless of ulcer depth. Finally, a higher percentage of women has controlled HbA1c levels (19.41% versus 11.95% in men) and more women in this group displayed non-infected wounds (60.6% and 43.90% for women and men, respectively).

Conclusion
Our results provide an updated picture of the DFI patterns and antibiotics resistance in patients attending Dasman Diabetes Institute (DDI) clinics in Kuwait which might help in adopting the appropriate treatment of infected foot and improving clinical outcomes. 1 Introduction outpatient center helping patients with diabetes mellitus manage their blood glucose levels and treat diabetes complications. Diabetic patients with complications were referred to DDI from all governorates in Kuwait and might be considered representative of all Kuwait population. During this period, 835 patients with DFU were referred to the Podiatry department, this represented around 120 patients per year. Microbiology samples were taken from 513 patients within this cohort at their first visit to our podiatry services. We obtained information on patients' age, sex, diabetes control, and treatment history of diabetes from the DDI Laboratory Information System (LIS) and the main parameters are summarized in Table 1. Approval and a waiver from the need to provide written informed consent were obtained from The Ethical Review Board of DDI for the access and use of anonymized data from the LIS for the purpose of publication.

Wound grading
Our experienced podiatrists classified the DFU according to the University of Texas Wound (UoT) Grading Classification System which grades DFU by depth (0, I, II, III) and stage (A, B, C, D) depending on the presence or absence of infection and ischemia [24]. This comprehensive and validated classification tool was consistently used for all subjects to develop a management plan and determine the eventual outcome of the DFU which are known to deteriorate with increased stages and grades. Clinical diagnosis of infection which has previously been defined [9,23] by the presence of at least 2 of the following indicators: local swelling or induration, >0.5 cm of erythema around the wound, local tenderness or pain, local warmth, and purulent discharge was used.

Samples collection and processing
Samples were taken by the podiatry staff at the patient's first appointment as per DDI policy at the time, or when there was clinical suspicion of DFI. Prior to sample collection the DFU was cleansed with sterile saline solution, and sharp debrided. Samples were collected either using a surface swab or wound biopsy. Surface swabs were collected using Levine's Technique [25], involving the rotation of a wound swab over a 1 cm 2 area of the wound for 5 seconds, using sufficient pressure to extract fluid from the inner part of the wound. Deep tissue specimens of 4mm in diameter were obtained from the base of the ulcer during the sharp debridement process. The specimens were placed into sterile transport containers and sent to the microbiology laboratory for aerobic culturing within 20 minutes. Anaerobic culturing was not performed in this study. Different types of media were used for sample inoculations; namely, blood, MacConkey, chocolate, mannitol-salt (MSA), Sabouraud agars, and thioglycolate broth (Oxoid, Basingstoke, UK). All plates were incubated for a maximum of five days or until visible growth using aerobic (O 2 ) at 37˚C, which ever come first; except for MRSA plates, which were incubated at 32˚C. Cultures with any volume of bacterial growth present on at least one of the inoculated plates was reported as laboratory positive. We did not consider lower or upper bacterial growth limits when reporting the presence of bacterial growth.

Identification of isolated microorganisms
All microbiological analyses were performed in our College of American Pathologist-(CAP) accredited Clinical lab at DDI (CAP Lab #8086466). Cultures were processed following the same CAP approved standard procedures for both swab and tissue samples and identification was done according to the Clinical & Laboratory Standards Institute (CLSI -M50 and CLSI-M100). All plates were examined for visible growth after 24 hours of incubation. Subcultures for thioglycolated broth tubes were done after a 24-hour incubation period, using blood, MacConkey, and Sabouraud agar plates. Wound smears from each sample were Gram-stained to identify if Gram-positive or Gram-negative isolates were present. We considered no-growth plates as sterile after five days of incubation. For positive swab or tissue cultures, we performed identifications using an automated Microscan system (Walkaway 40 SI, Siemens Healthcare Diagnostics, and Sacramento, CA). We obtained panels for Gram-negative (NC34, NC50 and NC53) and for Gram-positive bacteria (PC21, PC39) from Siemens Healthcare Diagnostics (Sacramento, CA). For confirmation, we performed further biochemical tests for both Grampositive and Gram-negative isolates using commercially available reagents (API 20E, API strep, and API staph) supplied by BioMérieux (Durham, NC, USA

Antimicrobial susceptibility testing
Susceptibilities to selected antibiotics for the common isolated bacteria (Staphylococcus aureus, Pseudomonas aeruginosa, Klebsiella pneumoniae, Escherichia coli, Acinetobacter sp and Enterobacter sp) were examined. Antimicrobial sensitivity testing of all isolates was performed on diagnostic sensitivity test plates according to the Kirby-Bauer method [26] following the definition of the Clinical and Laboratory Standards Institute (CLSI, 2014). Antibiotics and their concentrations used for antimicrobial susceptibility testing, are as following: Amikacin 30ug, Ampicillin10ug, Co-Amoxi/clav30 ug, Ceftazidime 30ug, Ciprofloxacin 5ug, Clindamycin 2ug, Ceftriaxone 30ug, Cefotaxime 30ug, Erythromycin 15ug, Gentamicin 10ug, Imipenem 10ug, Linezolid 30ug, Meropenem 10ug, Piperacillin/Tazobactam 110ug, Trimethoprim-Sulfamethoxazole 25ug, and Vancomycin 30ug. Bacterial inoculums were prepared by suspending the freshly grown bacteria in 5mL sterile saline solution. A sterile cotton swab was used to streak the surface of Mueller Hinton agar plates (Oxoid, Basingstoke, UK). Filter paper disks containing a designated concentration of the antimicrobial drugs obtained from Becton and Dickinson Company (Franklin Lakes, NJ) were used.

Statistical analysis
We analyzed the data using the statistical SPSS software, v22.0 (IBM SPSS Statistics for Windows, IBM Corp. Armonk, NY, USA). We evaluated differences between continuous and categorical variables using the Student's t-test and the chi-square with F-test, respectively. We calculated correlations between variables with the Spearman's rank correlation test. All statistical assessments were two sided and considered significant at p values <0.05.
Using the University of Texas (UoT) Wound Classification System, during the podiatry appointment, at the time of sample collection, we classified 253 (49.3%) DFU as clinically infected. This included 30.4% as UoT grade B (infection only) and 18.9% as UoT grade D (infection and ischemia) from the total study population. The remaining 260 patients (50.7%) had no clinical signs or symptoms of infection, including 42.5% as grade A and only 8.2% as grade C (Table 2). Within the whole study population just over one-quarter (27.1%) of ulcers were complicated with ischaemia, of which 69.7% also had infection. There was no difference in sex between the two groups, with neither sex being more likely to present with an infected ulcers than the other (49.6% and 48.8% of males and female, respectively; p = 0.642) ( Table 2 & Fig 1A). Similarly, there was no difference in means between age (p = 0.650) or HbA1c (p = 0.310). In addition, when splitting the two groups depending on their controlled and uncontrolled HbA1c levels, no significant difference was observed between the groups (Fig 1A & 1B). Nevertheless, there was significantly more males with deeper and more complex DFU than females (χ2(2) = 6.135; p = 0.047). Interestingly, there was a significant difference between the mean age and ulcer classification; patients with infected ulcers were approximately 5 years younger than patients with infection and ischemia (60.3±10.7 and 65.9±13.4 respectively, p <0.001). Additionally, this trend was continued when we took into consideration depth of DFU (p = 0.002). There were 207 clinically uninfected (Grade A & C) and 196 clinically infected (Grade B & D) DFU with HbA1c levels below 10.0% and 53 uninfected and 57 infected FDUs with HbA1c greater than 10.0%. However, there was no significant difference in the distribution of HbA1c levels between infected and uninfected DFU classification (p = 0.455) ( Table 2). Using Chisquare analysis, no significant association was observed between diabetes control and the presence of infection (χ2(2) = 0.3503; p = 0.554). Nevertheless, when we grouped the infected and uninfected DFU by UoT depth, a significant difference in mean HbA1c was found (F(5, 246) = 2.584, p = 0.027). Nearly 51% of DFU in our patients had no clinical signs of infection and were graded as UoT A or C. Interestingly, 47.7% of grade A and 42.9% of grade C DFU grew at least one of the following bacteria, Methicillin Sensitive Staphylococcus aureus (MSSA), MRSA, and Pseudomonas aeruginosa (data not shown).

Microbial isolate etiologies in ulcers with clinical infection
To better characterize the microbial etiology related to DFI, in the remaining part of our study and analyses we only included the clinically infected wounds (n = 253). Tables 3 and 4 summarize the distribution of microorganism isolates among the study population with clinical infection. Ninety-two percent of all samples collected grew one or more isolates. In these patients with infected ulcers, the mean age was significantly higher in those without growth (66.4 ± 9.2; CI, 63.6-69.3 years) than those with bacterial growth (61.3 ± 10.9; CI, 60.3-63.3 years); t(53.2) = −2.96, p = 0.005. However, we found no such significance in sex or HbA1c between those two groups (p = 0.108). Most microbiological samples were
We further analyzed the number of isolates present in those ulcers with clinical infection but different depths (I, II, III), however, no association was found between these parameters (χ2(7) = 6.27; p = 0.180). In contrast, Staphylococcus aureus and MRSA were predominantly found in infected ulcers while Pseudomonas aeruginosa was predominantly found in ulcers with infection and ischemia as reflected by the number of patients in each of the DFU grades (Table 4) and the relative distribution of these 3 strains within each of those grades (Fig 2) (p<0.05). Indeed, Staphylococcus aureus and MRSA were isolated from 66.1% and 85.7% of

PLOS ONE
infected ulcers as compared to 33.9% and 14.3% respectively in DFU with infection and ischemia, respectively, whereas Pseudomonas aeruginosa was present 61.1% of DFU with infection and ischemia compared to only 38.9% in infected ulcers (Table 4). In the next step, we analyzed the potential association between bacterial strain and depth (I, II, III). DFU penetrating to bone or joint (BIII or DIII) had a higher proportion of MRSA in comparison to DFU with less depth, however, the limited number of samples in some sub-groups did not allow us to draw a clear conclusion on the association of bacterial strains within DFU subgrades (Fig 2).

Discussion
DFU is expected to become a major diabetes-associated problem in Kuwait, as in many other countries, due to the high prevalence of diabetes in the population. However, only limited data on the subject is available in the country [13,15]. In this paper, we report estimates of DFU classification and types of DFU infection from our diabetes-dedicated institute for a period of 7 years. Our main findings were: (i) DFU was twice more likely in men than in women in our clinics, and more often present in younger men than in women; (ii) despite being younger, more males present with deeper and complex DFU than females; (iii) only about half of the   DFU were clinically infected (49.3%) but 92% of DFU showed bacterial growth in the microbiological lab analysis; (iv) comparable global findings of Gram-positive and Gram-negative strains, but significantly more Gram-positive strains are present in DFU without ischemia while more Gram-negative strains are present in DFU with ischemia; and (v) while Staphylococcus aureus was common in infected DFU without ischemia, Pseudomonas aeruginosa was more predominant in DFU with infection and ischemia, regardless of ulcer depth.
Our study population contained twice as many men as women with DFI, yet, our clinics have equal access to both sexes. This may be partially explained by a previous report which found that male sex and poor glycemic control are independent risk factors for DFI [27]. Similar trends have been observed in other countries, and authors have suggested that men are more likely to work outdoor which ultimately increases the risk to foot trauma and injury [28]. In addition, DFU were more often present in younger men than in women and despite being younger, more males presented with deeper and more complex DFU than females. Furthermore, we have observed an increased percentage of DFI in men at earlier ages than in women even without reaching statistical significance. This may be explained by the tendency of women to take more responsibility for medical care and personal hygiene, even though other factors cannot be ruled out. This hypothesis is further supported by our observations that (i) a higher percentage of women was present in the controlled HbA1c group (19.41% versus 11.95% in men) and (ii) more women in the controlled HbA1c group displayed non-infected wounds; grades A and C (43.90% and 60.6% for men and women, respectively) ( Table 2 and Fig 1A).
Globally, the number of isolates found in DFI varies widely. Polymicrobial infection rates have been found to be more prevalent in studies from a number of different countries including 83.7% in Portugal [29], 60% in South India [19], 55.7% in Mumbai, India [30], and 40% in Italy [31], whereas others have reported higher rates of monomicrobial infections [32,33]. Nevertheless, this variation in prevalence could be also related to sampling and processing techniques in each study. Overall, we found a higher percentage of DFI with monomicrobial isolates (57.3%) compared to polymicrobial isolates (34.8%) (1.30 pathogens per patient on average) and 8.2% with no growth. After excluding the no-growth samples, the average pathogens per patient increased to 1.43. Conversely, in Kuwait, two older hospital-based studies have reported high rates of polymicrobial infection (75% and 64%) in studies including 440 [15] and 86 [13] patients with DFI. This shift in number of isolates may be attributed to changes in DFU management, through regular follow-up and better patient education and knowledge throughout the country in the recent year. In Addition, DFI microbial etiology can change due to variations in healthcare systems and standard protocols as well as ethnicity and health status of the patients studied. In our study, and in agreement with a previous report from a teaching hospital in Kuwait [15], both Gram-positive and Gram-negative pathogens were similarly frequent in clinically infected DFU and no difference was observed in this pattern between sexes.
In the initial microbiological analysis of our DFI patients, we included both those with and without clinical infection as per the practices we had at the time in our institute. About half did not present with clinical signs of infection, these findings are similar to a previous large multicenter study of 1229 new DFU which reported an infection rate of 58% [34]. When performing microbiology culture analysis in our non-clinically infected patients, around 45% showed bacterial growth including at least one of the following bacteria, Staphylococcus aureus, MRSA, and Pseudomonas aeruginosa. Nevertheless, this bacterial presence was considered as colonization rather than infection. It is worth noting that following our current study and other unpublished data, the procedure has changed in our institute and now cultures are only collected from DFU with clinical signs of infection. This change in procedure is expected to reduce the costs involved in the management of DFU.
Interestingly, in our current analysis we found a significant trend between DFU depth and bacteria groups, where superficial infected DFU not involving tendon, capsule or bone and DFU penetrating to tendon or capsule were more likely to be infected by Gram-positive bacteria than those of the same depth and further complicated by the presence of ischemia. Accordingly, Staphylococcus aureus and MRSA were predominantly found in grades B while Pseudomonas aeruginosa was predominant in grade D wounds with underlying peripheral arterial disease. No association was, however, found between the number of isolates present and the DFU grade. We further assessed the potential association between DFU depth and the type of pathogens. However, even though DFU penetrating to bone and joint had a higher proportion of MRSA in comparison to wounds with less depth, the limited number of samples in some sub-groups did not allow us to draw a clear conclusion on this key question. In this context, different bacterial profiles have been reported with different wound grades, in which aggravated wounds and infections were associated with increased Gram-negative species, especially Pseudomonas aeruginosa [35]. However, debate remains open regarding the impact of polymicrobial infections on wound severity, as they have been reported to be more common in moderate wounds [35], while other reports suggest that polymicrobial infection increased in line with wound grading [36]. In a follow-up study over 10 weeks, Gardiner et al. examined DFI microbiome and found that it differed significantly and had a reduced diversity compared to that of control non-DFU wounds, and that this difference was not driven by the most abundant members of the wound pathogen community [37]. Those results and others, including our own, highlight the need for further detailed and goal-oriented designed studies to establish the potential association between strains and DFU grading before translating these debatable findings to clinical practice.
The observation that Staphylococcus aureus is the most common pathogen (19.9% excluding 5.0% MRSA) in our DFU patients followed by coagulase-negative Staphylococcus (15.7%) is concordant with most previous studies, including those from Kuwait, which also reported a high frequency of Staphylococcus aureus in DFI. This latter species causes the majority of human infections and has proven to be among the most persistent pathogens in healthcare facilities worldwide [38]. The MRSA rate in study is also comparable to the range reported in other countries, although prevalence rates vary widely [39,40]. The reduction in polymicrobial infections in our study, in comparison to previous reports from Kuwait [13,15], may reflect an improvement of healthcare system and in patients' hygiene. However, we cannot rule out other potential reasons affecting this rate, such as the younger age of our studied population. It is noteworthy that in our study 2.5% of Staphylococcus aureus isolates have shown resistance to Linezolid. Despite most of Staphylococcus aureus isolates including MRSA, are susceptible to linezolid, the emergence of linezolid-resistant MRSA has been, however, previously reported [41,42]. While E. coli and Klebsiella species were 100% sensitive to Carbapenems, Pseudomonas aeruginosa was relatively less sensitive to this class of antibiotics (88.4% and 85% sensitivity, respectively to Meropenem and Imipenem). These results are comparable to a previous report from Kuwait educational hospital where Pseudomonas aeruginosa was 10% resistance to Imipenem [15].
On the other hand, the Gram-negative Pseudomonas aeruginosa was identified in 12.8% of cases as compared to previous reports from Kuwait (17.4% and 17.5%) [13,15] which appeared higher than those in other geographical regions [43]. Interestingly, Pseudomonas aeruginosa, a multi-drug-resistant organism, has been reported to be highly prevalent in DFI patients from India in many studies [19,30,44]. These observations on the prevalence of Pseudomonas aeruginosa suggest regional variability probably due to microbial environments and healthcare practices such as sample collection and processing. Finally, and after dividing the data (DFU classification or infection) according to individual HbA1c levels (controlled or uncontrolled) we could not identify specific trends or associations between HbA1c levels and types of isolates or wound grades, despite the higher rate of men with controlled HbA1c and infected wounds.
Altogether, and in the context of the alarming worldwide increase both in the prevalence of DFI and antimicrobial resistance, our results are in favor of accurate and prompt identification of DFI to ensure the responsible isolates are targeted at earliest with the appropriate antibiotics, thus limiting the unnecessary use of broad-spectrum empirical antibiotics.
Despite the relatively large number of patients and the long period of time covered in our study, some limitations need to be highlighted. One unavoidable pitfall is the fact that our samples are a mixture of swabs and tissue samples that may not give consistent results, in particular with regards to aerobic and anaerobic pathogens as we did not include any condition or analysis of the latter group. In this context, using whole genome sequencing (WGS) approach may provide a more accurate representation of microorganism profiles including non-culturable organisms and a better estimate of the microorganism "load" in DFU. Furthermore, in such studies, only DFU with clinical infection are usually investigated at the microbiological level whereas here we initially included both DFU with and without clinical infection. A standardized procedure will help improve the consistency of microbial quantitation. Indeed, the implementation of guidelines of the International Working Group on the Diabetic Foot for obtaining specimens for culture from patients with DFI has been shown to be have cost-saving potential and provided interesting quality indicators in the global management of DFI [45]. In that study, within a period of 4 years, there was a significant decrease in the number of bacterial species per sample (from 4.1 to 1.6) and in the number of MRSA cases (from 52.2% to 18.9%). Another limitation is the relatively small number of patients in some wound classification groups, which did not allow us to analyze the data to find associations between various pathogens and the wound grading. Finally, our study was retrospective and used only UoT DFU classification system, therefore, a follow-up cohort study with these patients is needed to shed light on potential prediction of clinical outcomes, including wound healing in relation with the type of pathogens, DFU classification, clinical management, and the control of diabetes.
In conclusion, we provided an updated picture of the DFI pattern and showed an improved DFI pattern in comparison to the results in the previous reports from Kuwait and the region. Clinicians should keep up to date on common pathogen patterns in their local area, due to vast regional variability especially when prescribing empirical antibiotics. A combined DFI grading with microbial profile and sensitivity tests will help guide clinicians on the appropriate treatment and wound management.
Supporting information S1