Spread of resistant gram negatives in a Sri Lankan intensive care unit

In many developing countries, improper use of antibiotics, poor sanitation and shortfalls in infection control facilitate the amplification and dissemination of resistant strains in health care institutes and the community [1]. Increasing air travel, global mobility, health tourism, animal exports and international trade facilitate the spread of resistant organisms to countries worldwide [2]. Therefore, antimicrobial resistance is a global crisis that no single country can fight against, in isolation.

The World Health Organization (WHO) in February 2017 published a list of priority organisms that needs new antibiotics urgently. These include carbapenem resistant Acinetobacter baumanii, carbapenem resistant Pseudomonas aeruginosa and enterobacteriaceae resistant to carbapenems and 3^rd generation cephalosporins at critical level, emphasizing the magnitude of the problems occurring due to these [3]. Gram-negative bacteria, including the above were also the most common cause of healthcare associated infections (HAIs) in many low and middle income countries, and the focus of this paper [4].

Antibiotic resistance is a major concern in the Sri Lankan health care sector, particularly in relation to HAIs, with higher rates of cephalosporin resistance than community isolates [5]. Carbapenem resistance is also known to be widespread among many institutions [6‐8].

Intensive care units (ICUs) have a high antibiotic pressure. Therefore, resistant organisms develop and thrive in these environments. High antibiotic resistance in ICU settings is present throughout Sri Lanka [7‐10]. While most studies describe the sensitivity patterns, mechanisms of resistance, genetic determinants of resistance and clonality have not been studied widely in Sri Lanka. Currently, surveillance also remains at the level of identification of organisms and resistance patterns via routine testing. Surveillance is considered as an integral part of infection prevention and control programmes [11]. Effective surveillance programmes are known to reduce rates of hospital acquired infections [12]. Despite the recent improvement of laboratory infra-structure in Sri Lanka, no routine typing protocols are in place. While methods such as Next Generation Sequencing (NGS) based typing and Multi Locus Sequence typing (MLST) have now become the standard methods in typing, in resource limited settings, simpler methods such as random amplification of polymorphic DNA (RAPD) can be used more cost-effectively.

This study was carried out in order to identify the feasibility of using RAPD and minimal clinical data for laboratory surveillance in resource limited settings.

This laboratory based study was conducted on consecutive Gram negative isolates obtained from cultured respiratory tract specimens received from patients admitted to the ICU of Teaching Hospital Peradeniya (THP) from March to September 2015. The study unit is the single ICU providing intensive care facilities to this 924 bed hospital and has 10 ICU beds that are housed in three compartments. The first isolate obtained per species was selected per patient unless there was a gap of >5 days between the two specimens or a difference in the morphological appearance or antibiogram. Ethical clearance for the study was obtained from the Institutional Ethical Review Committee, Faculty of Medicine, University of Peradeniya, Sri Lanka.

Isolates were stored at −80 °C prior to testing. Date of admission to the ICU and date of discharge were obtained retrospectively from the institutional records.

All isolates were identified up to their species level except the Acinetobacter spp., by a using series of standard biochemical tests supplemented by commercial kits where needed. Acinetobacter spp. were identified up to genus level.

Antibiotic susceptibility testing (ABST) was performed and interpreted accordingly to the Clinical and Laboratory Standards Institute (CLSI) guidelines and standards (CLSI, 2016). ESBL production was confirmed in E. coli and K. pneumoniae isolates by the combined disc test [13]. Carbapenem non-susceptible (either resistant or intermediate sensitive) E.coli and K. pneumoniae isolates were tested for carbpenemase production by the modified Hodge test (MHT).

As genetic determinants of ESBL production, bla _CTX-M, bla _SHV, and bla _TEM genes were detected using previously described primers in all E. coli and K. pneumoniae isolates [14]. PCR reactions were carried out with 2 μL of template obtained from boil lysis of organisms, 5 μL of 5× Green GoTaq™ buffer (Promega, USA), 3 μL of 25 mM MgCl₂ (Promega, USA)_, 1 μL of 10 mM dNTP mix (Promega, USA), 0.4 μL of 5 U/μL GoTaq™ Flexi DNA polymerase (Promega, USA) and 0.5 μL of each 10 μM forward and reverse primers (Integrated DNA Technologies) with molecular grade water (Invitrogen, USA) added to a total of 25 μL. The annealing temperature for the ESBL multiplex PCR was 60 °C.

Six primers were screened using DNA extracted from K. pneumoniae BAA 1705, E. coli ATCC 25922, Pseudomonas aeruginosa ATCC 27853 and a clinical Acinetobacter spp. isolate (isolate no 105). The primer giving rise to the highest number of discrete bands was selected for typing the given group of organisms. Selected primers were RAPD_1:GAAGCAGCCCGGTAGTAGGTTGAG for E. coli and K. pneumoniae [15], RAPD_5: AGCGGGCCAA for Acinetobacter species and RAPD_6: ACGGCCGACC for Pseudomonas aeruginosa [16]. The PCR mater-mix constituted of 3 μl of 25 nM MgCl₂ (Promega, USA), 5 μl of 5× Green GoTaq Buffer (Promega, USA), 1 μl of 10 mM dNTP mix (Promega, USA), 0.3 μl 5u/ μl of GoTaq Flexi DNA polymerase, 2 μl of 10 mM primer (Integrated DNA Technologies), 2 μl of template DNA and 11.7 μl of molecular grade water (Invitrogen, USA) to a total volume of 25 μl. All reactions were performed at an annealing temperature of 36 °C for 1 min.

All gel pictures captured were analyzed with the GelJ software downloaded from http://​sourceforge.​net/​projects/​gelj/​ on 30/01/2016 [17]. Dendograms were drawn with Dice co-efficient for similarity calculations using UPGMA linkage. Tolerance levels were adjusted for the given species of isolates with the reference lanes.

During the six month study period a total number of 70 Gram negative isolates were obtained from respiratory tract specimens. Of these 70 isolates, seven isolates were E. coli (10%), 14 were K. pneumoniae (20%), three isolates were Stenotrophomonas maltophilia (4.3%) and one was Aeromonas hydrophila (1.4%). Also 15 of the 70 isolates were identified as P. aeruginosa (21.4%) and 30 were identified as Acinetobacter spp. (42.9%).

This study identified 25 coliforms (35.7%) out of 70 Gram negative isolates. All seven E.coli and all 14 K.pneumoniae isolates were found to be ESBL producers. A study on blood stream infections in Sri Lanka had identified that 22.5% of adult septic patients had ESBL producing organisms as causative agents [10]. Similarly, in a study conducted on urinary isolates, 40.2% had been identified as ESBL producing enterobacteriaceae [19]. Both these studies would have included more patients with community onset disease than the cohort of the current study, accounting for the differences in resistance rates. Considering mainly health care associated infections, a recent study performed in an Indian tertiary hospital founded that among 223 isolates of K. pneumoniae, 114 (51.1%) were ESBL producers [20]. A long term study at an Indian institute had identified ESBL prevalence to be as high as 80.9% within certain ICUs [21].

All E.coli isolates were found to harbour bla _CTX-M. CTX-M has emerged to be the most common genetic determinant for ESBL production globally, including Sri Lanka [19, 22]. The most common genetic determinant of ESBL production among the K.pneumoniae isolates was blaSHV, followed by blaCTX-M. A separate study in Sri Lanka has also found a higher presence of blaSHV among K.pneumoniae isolates than E.coli isolates [19]. Similar findings have been reported from elsewhere in the world too [23]. A number of K.pneunoniae isolates (50%) were found to carry all three types of ESBL associated genes. It is alarming to notice this trend, particularly as the genetic determinants of resistance may be found on transmissible genetic elements.

The available therapeutic options for the treatment of ESBL-associated infections are limited due to co-resistance to various groups of antibiotics. Carbapenems are the the drug of choice against ESBL producers. However, with the increasing rates of ESBL producers, carbapenems are being used increasingly as empirical therapy. This in-turn drives the development of resistance to carbapenems. In this study six Klebsiella pneumoniae isolates (43%) were found to be non-susceptible for carbapenems and were confirmed to be carbapenemase producers by the MHT. A Sri Lankan study conducted from January to April 2012 had identified a carbapenamase production rates of 7.9% and 0% at two institutes in the Western province of Sri Lanka [6]. However, this study included isolates other than those obtained from ICUs.

In our study, which was conducted at a center over a period of 6 months, the majority of respiratory tract isolates were Acinetobacter spp. (n = 30, 43%). In a study conducted in 15 ICUs across Vietnam, the most frequently identified hospital associated infection related isolates were A. baumannii (n = 726, 24.4%) [24]. In a recent study conducted in Sri Lanka, out of 200 patients with VAP, 33.5% of the patient samples yielded Acinetobacter species [7]. Our study included organisms isolated from patients, irrespective of whether the patients had underlying Ventilator Associated Pneumonia (VAP) or not and some of the isolate could have been colonizers. However, if a patient develops subsequent VAP, it is likely that the colonizing flora would give rise to the infections; therefore, our sample selection could be justified coupled with the difficulty in establishing the diagnosis of VAP [25]. Furthermore, a recent long term study in the United States has identified that the presence of carbapenem resistant A. baumannii on surveillance cultures was strongly associated with the later development of infections with the same organisms, further validating our selection of samples [26].

In our study, all seven of the E. coli isolates, all 14 Klebsiella pneumoniae isolates and all 30 Acinetobacter spp. isolates could be defined as MDR as they were resistant to more than three antimicrobial classes. Out of the 15 P. aeruginosa isolates, only two MDR isolates (13%) were found to be MDRs. A cross sectional study which was conducted in Siriraj Hospital, Thailand considering all hospitalized patients who had positive culture for enterobacteriaceae, P. aeruginosa and A. baumannii during February to May 2012 revealed 48.8% of prevalence in overall MDR gram-negative bacteria. The percentage of MDR organism was 37.8% for ESBL producing enterobacteriaceae, 39.3% for carbapenem-resistant P. aeruginosa and 88.7% for MDR A. baumannii [27]. Our figures are somewhat higher than this study, which included patients from all units. The emergence of MDR Gram negatives in the ICUs is a major concern and needs urgent interventions in a country wide scale. As a number of genetic determinants of resistance can occur within the same mobile genetic elements (MGE), it would be interesting to identify the possible MGEs in our isolates.

The RAPD analysis of the four types of organisms studied revealed that in all types, a predominant cluster could be identified, rather than multiple clusters or singletons. This hints towards a possible within unit transfer of organisms rather than multiple acquisitions from outside environments. Similarly, graphical representations of the duration of hospitalization among the patients also demonstrate that there is overlap among the stay of patients carrying similar organisms belonging to given RAPD clusters as well as sporadic introduction of other cluster. Environment could be a potential source for the spread of the same cluster between patients who did not have an overlapping stay. These highlight the importance of adhering to proper infection control protocols within infection care units. Further analysis of isolates for a longer period of time would enable a time-series analysis of the different clusters.

While RAPD is a simple typing method, a typing method with more resolution might have given a better resolution that may have enabled a better differentiation of the strains. Performing a typing method such as MLST on selected RAPD clusters would strengthen the current results. However, the ability to perform RAPD in a relatively cost effective manner and the ability to analyze the fingerprint data using a free software that is easy to use with a graphical user interface, demonstrates that this method could be used in low cost settings. RAPD has also been integrated successfully in infection control programmes elsewhere [28].

There were several limitations in the current study. Our study was performed in a single center over a relatively shorter period. Therefore, time trends in resistance cannot be assessed. We also did not differentiate colonizing agents from agents causing infections. Genetic determinants of resistance was identified only among the enterobacteriaceae and only for ESBL production. Further studies need to look in to carbapenamase producers as well. The sensitivities were conducted using disc diffusion testing and minimum inhibitory concentrations (MICs) were not performed. Furthermore, we included isolates from respiratory specimen only. We also did not analyze the details in relation to the underlying conditions warranting ICU admissions. However, by using RAPD and hospitalization timelines, we could demonstrate the possible linkage of organisms included in the study.

In conclusion, simple RAPD based typing followed by graphical analysis of hospitalization data can be used to identify trends in the presence and spread of resistance organisms in a cost effective manner. Data obtained in this manner could be a more convincing tool to persuade health-care staff to adhere to infection control protocols. Such surveillance could be made an integral part of the infection control programmes and could be used in possible outbreak situations to evaluate the spread and effectiveness of control measures.