Epidemiology of invasive infections caused by vancomycin sensitive and resistant enterococcal strains among oncology patients at the National Cancer Institute of Sri Lanka from 1st of July 2012 to 31st of July 2013

Introduction and Objectives: Enterococci have traditionally been regarded as low grade pathogens but have emerged as an increasingly important cause of nosocomial infections in the last decade. While the Enterococcus faecalis remains the predominate species in clinical infection, Enterococcus faecium isolates are increasing in proportion. This study was carried out to describe the epidemiology of invasive enterococcal infections among oncology patients at the National Cancer Institute of Sri Lanka (NCISL). Methods: 60 patients with invasive enterococcal infections, who were treated as inward patients at the National Cancer Institute of Sri Lanka from 1.7.2012 to 31.7 21013 whose samples were sent for microbiological investigation were included in this study. Speciation of the isolates was done by using a rapid manual analytic system (RapID STR panel-Oxoid). Vancomycin sensitivity was assessed in all enterococcal isolates by disc diffusion method, agar dilution screening method (CLSI guideline 2013) and detection of minimum inhibitory concentration (MIC) by using Vancomycin gradient strips. Teicoplanin MIC was assessed only in vancomycin resistant isolates using teicoplanin gradient strips. Associated factors for getting an infection with vancomycin resistant strains were assessed using a data extraction sheet. Results: The incidence of invasive enterococcal infections among oncology patients was 1.1 per 1000 admissions. The incidence of invasive enterococcal infection caused by vancomycin resistant species was 0.16 per 1000 admissions. E. faecium was the dominant species causing invasive enterococcal infections (55%). Using the gradient strip method for determination of vancomycin MIC as the gold standard, the screening agar dilution method had 100% sensitivity and specificity and the disc diffusion test had a low specificity (77.8%). Almost all the participants (96.6%) had acquired the enterococcal infection from the hospital. Vancomycin resistant infections were more common in patients who had haematological malignancies, who were treated with 3 generation cephalosporins and cytotoxic chemotherapy drugs.


Introduction and Objectives
Although about a dozen Enterococcus species have been identified, E. faecalis and E. faecium are responsible for most human infections. Historically, the ratio of infections due to E. faecalis to those due to all other Enterococcus species was approximately 10:1. 1 In recent years, there has been a progressive decline in this ratio. This microbiological shift is likely to be explained in part by the emergence of vancomycin resistant enterococci (VRE) and E. faecium being the dominant species identified among VRE. 2 The most common nosocomial infections caused by these organisms are urinary tract infections followed by intra-abdominal infections, pelvic infections, surgical wound infections, bacteraemia, endocarditis, neonatal sepsis and rarely meningitis. A major reason why these organisms survive in the hospital environment is due to a high level of intrinsic resistance to many antibiotic groups (ß lactams, low concentrations of aminoglycosides, nalidixic acid, clindamycin, fluoroquinolones and trimethoprin-sulfamethoxazole). 3 They are also able to acquire resistance to many commonly used antibiotics (glycopeptides, tetracyclines, erythromycin, fluoroquinolones, rifampicin, chloramphenicol, fusidic acid, nitrofurantoin). 4,5 Vancomycin resistant enterococci are classified according to their phenotype using vancomycin and teicoplanin MIC 6 and genotype using the specific ligase gene. 7 There are 5 recognized phenotypes (VanA, VanB, VanC, VanD, VanE) and 7 genotypes (VanA, VanB, VanC, VanD, VanE, VanG, VanL). Motile enterococci, E. gallinarum and E. casseliflavus have an intrinsically low level of resistance to vancomycin and belong to the VanD genotype. 7 These two species have been identified as a clinically significant cause of bacteraemia in immunocompromised patients. 8 Most of the risk factors for infection with VRE, namely immunosuppression, advanced age, other co-morbidities such as renal and liver failure, invasive procedures and devices and gastrointestinal surgery are present in oncology patients. 7 Use of third generation cephalosporins, carbapenems and glycopeptides are also regarded as risk factors for acquisition of VRE and other glycopeptide resistant enterococci. 7 These antibiotics are frequently used as empirical treatment for febrile neutropenia in oncology patients (NCISL antibiotic policy). No study has been done in Sri Lanka to describe the epidemiology of vancomycin sensitive and resistant enterococcal infection in oncology patients or screening of patients to evaluate the prevalence of colonization with VRE. This study was therefore carried out to describe invasive enterococcal infections caused by vancomycin sensitive and resistant strains among oncology patients at the National Cancer Institute of Sri Lanka.
The objectives of the current study were to describe the following characteristics of Enterococcus spp. isolated from patients with invasive infections in hospitalized oncology patients at National Cancer Institute, Sri Lanka. a. Determining vancomycin sensitivity b. Speciation of isolates c. Identifying factors associated with VRE infections d. Assess teicoplanin sensitivity of VRE isolates.

Methods
This study was conducted as a descriptive cross sectional study from 1.7. 2012 to 31.7.2013.
Sixty patients with invasive enterococcal infections were included in this study. A case of invasive enterococcal infection was defined by the isolation of Enterococcus spp. from a normally sterile body site. Sterile sites include blood, peritoneal aspirates, pus aspirates from abscesses and sterile body cavities, aspirate from biliary tract, pleural aspirates and cerebrospinal fluid. Urine isolates were also included if there was evidence of pyelonephritis. Nosocomial infection was defined as infection obtained >72 hr. after hospital admission.
The sample size of patients with invasive enterococcal infection necessary to obtain the required precision (0.3) and confidence (95%) level was 60.
A pilot study was done from 1.1.11 to 31.7.2011 using laboratory work sheets to determine the prevalence of invasive enterococcal infection in oncology patients at NCISL and prevalence was calculated as 0.7%. Using this prevalence, the required sample size of VRE to obtain a precision of 0.3 and confidence level of 95% was determined to be 60 which could be expected from a total sample size of 8570. During the study period 9652 samples were processed which included 4051 blood cultures, 4652 urine cultures, 855 pus aspirates, 35 pleural fluid samples, 36 peritoneal fluid samples and 23 cerebrospinal fluid samples.
Enterococcal blood stream infection was defined as the isolation of an Enterococcus spp from one peripheral blood culture sample. When blood cultures from both central catheter and peripheral site were collected at the same time, if blood culture through the catheter became positive two or more hours before peripheral blood culture, it was considered as catheter associated blood stream infection. Enterococcal bacteremia occurring >60days after a previous episode was considered to be a separate blood stream infection. In urine cultures, a pure growth of ≥10 5 colony forming units (CFU)/ml in patients with symptoms suggestive of pyelonephritis (fever with chills and rigors with loin tenderness) or patients with ultra sound scan evidence suggestive of pyelonephritis only were included in this study.
Gram stain was performed on colonies which had growth on both blood and MacConkey agar (without salt) and which were catalase negative. In urine cultures, Gram stain was performed on lactose fermenting catalase negative colonies on CLED medium. Gram positive cocci in chains were Lancefield grouped. Lancefield group D isolates were further tested for growth on 6.5%. NaCl, survival at 60  C for 30 minutes, and hydrolyzing of L-pyrrolidonyl-b-napthylamide (PYR) to identify enterococci. All isolates which showed growth on 6.5% NaCl, a positive result for PYR and survived 30 minutes at 60  C were included in the study. 9 Speciation of isolates was done using a rapid manual analytic system (Rapid STR panel-Oxoid) according to the manufacture instructions. Rap ID STR panels contain 10 reaction cavities for different biochemicals and test cavities seven through ten are bifunctional. In addition to these biochemical tests, haemolysis on blood agar plate was assessed according to the manufacturer's instructions. The microcode obtained from the test kit was compared with the rapid STR code compendium for the identification. Evaluation of motility and pigment production was also done to identify motile and pigment producing enterococci.
Associated factors for getting an infection with VRE, namely prior administration of antibiotics and chemotherapy, invasive procedures and devices, gastrointestinal surgery, renal and liver failure, diabetes mellitus and mucositis, were assessed using a data extraction sheet. Data was obtained by reviewing medical records of each study patient.
Three methods, disk diffusion method using 30 µg vancomycin discs (Oxoid), agar dilution screening method and minimum inhibitory concentration (MIC) determination using gradient strip method (Oxoid vancomycin MIC Evaluator) were used to detect vancomycin susceptibility .The gradient strip method (Oxoid teicoplanin MIC evaluator) were used to detect MIC of teicoplanin on vancomycin resistant isolates.The disk diffusion test and the agar dilution screening methods were performed according to the recommendations of performance standards for antimicrobial susceptibility testing, clinical laboratory standard institute, M100 ,2013 (CLSI, 2013). 10 According to the CLSI guideline vancomycin zone diameter criteria were ≥17mmsensitive, 15 To describe the phenotype of VRE, phenotypic classification originally described by Gold et al. 11 was used. VanA isolates have high MICs of vancomycin (MIC range 64-1,000 µg/ml) and teicoplanin (MIC range 16-512 µg/ml), whereas VanB isolates often have lower MICs of vancomycin (MIC range 4-1,024µg/ml but usually on the low side) but typically susceptible to teicoplanin. Van D isolates have moderate susceptibility to both glycopeptides (vancomycin MIC range 2-32; teicoplanin MIC usually <0.5µg/ml ) whereas VanC and VanE isolates display low level resistance to vancomycin (MIC range 0.5-4µg/ml) and are susceptibility to teicoplanin.

Results
Of the 9652 samples, there were 60 clinically and microbiologically diagnosed patients with invasive enterococcal infections.
Mean age of the participants who were in the pediatric age group and non-pediatric age group are 4.92 years (SD=3.75) and 47.7 years (SD = 18.47) respectively ( Table 1).
During the period of this study there were 12688 new registration of patients and total number of admissions was 54027.The incidence of invasive enterococcal infections among oncology patients was 1.1 per 1000 admissions. The incidence of invasive enterococcal infection caused by vancomycin resistant spp. was 0.16 per 1000 admissions.
Almost all the participants (96.6%) had acquired the enterococcal infection from the hospital ( Table 2). The majority of the participants were admitted in the medical wards at the time of the positive culture. The highest proportion of vancomycin resistant infections were from pediatric ICU (23.3%) followed by medical wards (17.2%) and all were hospital acquired infections ( Table 3).   Enterococci were isolated mainly from patients with UTI and with bacteraemia, with E. faecalis the predominant species isolated from these 2 sites as well as from pus aspirates (   Of the invasive enterococcal infections in patients with haematological malignancies, 22.7% were caused by resistant strains, compared to only 10.5% among patient with solid tumors. 16% of patients who had been given 3 rd generation cephalosporins had infections with resistant strains, while resistant strains were not isolated from patients who not treated with a cephalosporin. Around 22% of patients who had been given chemotherapy had infections with resistant strains. However, these findings were not statistically significant ( Table 7).
Only 2 17 of patients with positive blood cultures complied with the necessary criteria to be identified as bacteraemia associated with an intravascular catheter. Vancomycin resistant isolates were not found in these 2 isolates although prevalence of vancomycin resistance among the patients who had a CVC was 22.2%.
Seven of the 9 patients with VRE had expired at the time of collecting data from medical records.
Using the gradient strip method as the standard, the disc diffusion method did not detect vancomycin resistance in 2 isolates, E. faecium and E.casseliflavus, with an MIC of 6μg/ml (Table  8 and 9). Using the same method, Vancomycin MIC of vancomycin sensitive strains ranged from 0.5-2 μg/ml. MICs of resistant strains is shown in

Discussion
A prospective study 12 in North India identified E. faecium as the commonest blood culture isolate while E. faecalis predominated in pus and urine samples. Other species isolated in this study were E. mundtii, E. dispar, E. durans, E. avium, E. raffinosus and E. gallinarum. In another study from New Delhi, E. faecium (66%) was the most common isolate in blood samples followed by E. faecalis (20%). 13 However, E. faecalis (55%) followed by E. casseiflavus (24%) and E. faecium (12%) were reported from Chandigarh from urinary isolates. 14 Perera et al. 15 at Sri Jayewardenepura General Hospital, Sri Lanka reported that isolation of E. faecium was 45% while E. faecalis remained the predominant species. In this study E. faecium was the predominant species which caused invasive enterococcal infections (55%) among oncology patients. There are no local or national surveillance data on enterococcal infections in Sri Lanka for comparison.  for E. faecium to become the predominant species and there is a need to conduct a proper case control study to identify the risk factors for VRE.
All the VRE infections were hospital acquired in this study population. However, the study was not designed to elicit an association between the duration of hospital stay and getting an infection with VRE.
The first vancomycin resistant enterococci (VRE) isolate that harboured the vanA transposon was identified in 1988 by Uttley et al 16 which was reported 30 years after vancomycin was clinically introduced. In 2003, 28.5% of enterococcal strains in the United States were vancomycin resistant. 17 In Australia, in 2005, VRE was low at 0.73%. 18 In 2007, 8.5% of enterococcal isolates in the United Kingdom were found to be vancomycin resistant. 19 Those reports included all enterococci isolates from all patients. In this study the incidence of vancomycin resistant enterococci causing invasive infections in oncology patients was 15%.
In a study evaluating the accuracy of eight currently available test methods (agar dilution, disk diffusion, Etest, agar screen plate, Vitek GPC-TA and GPS-101, and MicroScan overnight and rapid panels) it was shown that VanA VRE were detected by all methods but vanB VRE were often not detected by Vitek GPS-TA and MicroScan rapid (sensitivities 47% and 53% respectively 20 All methods except Etest and agar screen continue to show problems in the detection of VanC1 and Vanc2 VRE. 20 The sensitivity of the disc diffusion test in detecting VanA resistant was 100% but was lower in detection of VanB, VanC1 and vanC2 (93%,52% and 63% respectively). 20 MIC detection by the broth dilution method is considered the gold standard to determine antibiotic sensitivity/resistance according to CLSI guidelines 2013. 10 Due to practical difficulties in using the broth dilution method, the gradient vancomycin strip was used as the standard in this study. The detection of vancomycin resistance by disc diffusion test remained lower (77.8%) than the agar screening test (100%).
MICs of teicoplanin in vancomycin resistant isolates varied from 6-24 μg/ml making these isolates of intermediate sensitivity to teicoplanin. Of 8 vancomycin resistant E. faecium isolates 7 isolates were phenotypically compatible with VanA phenotype of glycopeptide resistant enterococci.
Of the 2 E. casseliflavus isolates, one was vancomycin susceptible according to the CLSI 2013 criteria. Since E. casselifavus isolates are intrinsically resistant to vancomycin, it is not recommended to treat infections with vancomycin even though the organism shows MICs in the sensitive range. 10 It is therefore important to speciate enterococcal species to detect such anomalies to prevent treatment failures.

Limitations
With this small sample size, precision of findings was low and power to detect the associations between risk factors and vancomycin sensitivity is low. A case control study is better to identify the risk factors.

Conclusions and recommendations
E. faecium was the dominant species (55%) causing invasive enterococcal infections in oncology patients at the National Cancer Institute of Sri Lanka in the study period (2012/13). 78% of vancomycin resistant E. faecium in the study were VanA phenotype of glycopeptide resistant enterococci. It is important to speciate enterococci to identify the intrinsically resistant Enterococci spp to prevent treatment failures by using inappropriate antibiotics.
Almost all the participants (96.6%) had acquired the enterococcal infection from the hospital. As most risk factors for VRE infections are common in oncology patients, it is recommended that surveillance of VRE is continued with inclusion of a case control study to identify risk factors.
The disc diffusion test was shown to have lower sensitivity in comparison with the agar dilution screening and E test. We therefore recommend one of the latter 2 tests in clinical laboratories investigating patients with invasive enterococcal infections.