Update of incidence and antimicrobial susceptibility trends of Escherichia coli and Klebsiella pneumoniae isolates from Chinese intra-abdominal infection patients

Background To evaluate in vitro susceptibilities of aerobic and facultative Gram-negative bacterial (GNB) isolates from intra-abdominal infections (IAIs) to 12 selected antimicrobials in Chinese hospitals from 2012 to 2014. Methods Hospital acquired (HA) and community acquired (CA) IAIs were collected from 21 centers in 16 Chinese cities. Extended spectrum beta-lactamase (ESBL) status and antimicrobial susceptibilities were determined at a central laboratory using CLSI broth microdilution and interpretive standards. Results From all isolated strains the Enterobacteriaceae (81.1%) Escherichia coli accounted for 45.4% and Klebsiella pneumoniae for 20.1%, followed by Enterobacter cloacae (5.2%), Proteus mirabilis (2.1%), Citrobacter freundii (1.8%), Enterobacter aerogenes (1.8%), Klebsiella oxytoca (1.4%), Morganella morganii (1.2%), Serratia marcescens (0.7%), Citrobacter koseri (0.3%), Proteus vulgaris (0.3%) and others (1.0%). Non- Enterobacteriaceae (18.9%) included Pseudomonas aeruginosa (9.8%), Acinetobacter baumannii (6.7%), Stenotrophomonas maltophilia (0.9%), Aeromonas hydrophila (0.4%) and others (1.1%). ESBL-screen positive Escherichia coli isolates (ESBL+) showed a decreasing trend from 67.5% in 2012 to 58.9% in 2014 of all Escherichia coli isolates and the percentage of ESBL+ Klebsiella pneumoniae isolates also decreased from 2012 through 2014 (40.4% to 26.6%), which was due to reduced percentages of ESBL+ isolates in HA IAIs for both bacteria. The overall susceptibilities of all 5160 IAI isolates were 87.53% to amikacin (AMK), 78.12% to piperacillin-tazobactam (TZP) 81.41% to imipenem (IMP) and 73.12% to ertapenem (ETP). The susceptibility of ESBL-screen positive Escherichia coli strains was 96.77%–98.8% to IPM, 91.26%–93.16% to ETP, 89.48%–92.75% to AMK and 84.86%–89.34% to TZP, while ESBL-screen positive Klebsiella pneumoniae strains were 70.56%–80.15% susceptible to ETP, 80.0%–87.5% to IPM, 83.82%–87.06% to AMK and 63.53%–68.38% to TZP within the three year study. Susceptibilities to all cephalosporins and fluoroquinolones were less than 50% beside 66.5% and 56.07% to cefoxitin (FOX) for ESBL+ Escherichia coli and Klebsiella pneumoniae strains respectively. Conclusions The total ESBL+ rates decreased in Escherichia coli and Klebsiella pneumoniae IAI isolates due to fewer prevalence in HA infections. IPM, ETP and AMK were the most effective antimicrobials against ESBL+ Escherichia coli and Klebsiella pneumoniae IAI isolates in 2012–2014 and a change of fluoroquinolone regimens for Chinese IAIs is recommended. Electronic supplementary material The online version of this article (10.1186/s12879-017-2873-z) contains supplementary material, which is available to authorized users.


(Continued from previous page)
Conclusions: The total ESBL+ rates decreased in Escherichia coli and Klebsiella pneumoniae IAI isolates due to fewer prevalence in HA infections. IPM, ETP and AMK were the most effective antimicrobials against ESBL+ Escherichia coli and Klebsiella pneumoniae IAI isolates in 2012-2014 and a change of fluoroquinolone regimens for Chinese IAIs is recommended.

Background
The Study for Monitoring Antimicrobial Resistance Trends (SMART) is a global surveillance program, which monitors annually in vitro antimicrobial susceptibilities of hospital acquired (HA) and community acquired (CA) intra-abdominal and urinary tract infections due to aerobic and facultative Gram-negative bacilli (GNB). Intra-abdominal infections (IAIs) are the second most common cause of sepsis in intensive care units (ICU) [1] where they are the second most common cause of infection-related mortality [2]. IAIs are also the second most common cause of infection related to surgical interventions and according to a multicenter observational study in 68 medical institutions worldwide, [3] the overall mortality rate of patients with complicated IAIs in 2012-2013 was 10.5%, [4], with ESBL producing bacteria being a particular challenge for treatment [5]. However, initiation of appropriate antimicrobial therapy can significantly reduce the mortality rate of IAI-induced septic shock [6]. Since appropriate antibiotic therapy is essential for IAIs [7,8], institutional and nationwide surveillance of IAI-derived bacterial strain susceptibilities provides crucial information for the selection of the right choice of empirical antimicrobial treatment.
Although a significant increase of the proportion of ESBL-positive Enterobacteriaceae hospital infections in Germany over the period 2007-2012 [9] and in Japan from 2000 to 2010 have been reported [10,11] the situation in China is not clear. A limited number of ESBLscreen positive Escherichia coli and Klebsiella pneumoniae IAI isolates from 2012 and 2013 have been documented, but there is a wide diversity in ESBL-related molecular characteristics [12].
The present study mainly focused on ESBL-screen positive rates of IAI isolates and concomitantly on resistance rates of IAIs, in particularly those caused by Enterobacteriaceae against, 3rd and 4th generation cephalosporins, a cephamycin, 2nd generation fluoroquinolones, carbapenems, an aminoglycoside, as well as a combination of drugs containing penicillins with β-lactamase inhibitors. The data was collected from 21 centers in 16 Chinese cities between 2012 and 2014 in order to provide guidance for antimicrobial therapy of IAIs.

Collection and identification of isolates
The Human Research Ethics Committee of Peking Union Medical College Hospital approved this study and waived the need for consent (Ethics Approval Number: S-K238).
GNB strains were collected from consecutive IAI patients between 2012 and 2014 in 21 centers located in 16 Chinese cities. Only gram-negative aerobic and facultative anaerobic bacteria from abdominal infection sites such as the appendix, peritoneum, colon, bile, pelvis and pancreas were included and the strains needed to be pathogenic bacteria associated with clinical infections while grampositive and anaerobic bacteria were excluded. The specimens were mainly obtained through surgical procedures, but puncture specimens such as intraperitoneal puncture fluid were also included and different gram-negative bacteria that were combined in one sample were also accepted. Exclusion criteria were isolates from drainage liquid or drainage bottles, as well as isolates from feces or perianal abscess environmental samples (not a patient source) or cultures for infection control purposes. All isolates were sent to the central clinical microbiology laboratory of Peking Union Medical College Hospital for initial bacteria identification and re-identification using MALDI-TOF MS (Vitek MS, BioMérieux, France). All organisms were considered clinically significant by local hospital criteria. Isolates collected within 48 h of hospitalization were categorized as CA IAIs, and those collected after 48 h were categorized as HA IAIs [13].

Extended-spectrum β-lactamase (ESBL) detection
Phenotypic identification of ESBL-screen positivity in Escherichia coli and Klebsiella pneumoniae (ESBL+) were carried out by CLSI recommended methods [15]. If cefotaxime or ceftazidime MICs were ≥2 μg/mL, the MICs of cefotaxime + clavulanic acid (4 μg/mL) or ceftazidime + clavulanic acid (4 μg/mL) were comparatively determined and ESBL production was defined as a ≥ 8-fold decrease in MICs for cefotaxime or ceftazidime tested in combination with clavulanic acid, compared to their MICs without clavulanic acid.

Statistical analysis
All of the statistical analyses were performed using the IBM SPSS Statistics for Windows (Version 19.0. Armonk, NY: IBM Corp). The susceptibility of all Gram-negative isolates combined was calculated using breakpoints appropriate for each species and assuming 0% susceptible for species with no breakpoints for any given drug. The 95% confidence intervals were calculated using the adjusted Wald method; linear trends in susceptibility and ESBL rates were assessed for statistical significance using the Cochran-Armitage test; P values <0.05 were considered to be statistically significant.
The relative percentages of ESBL-screen positive Escherichia coli and Klebsiella pneumoniae strains from IAI isolates showed a decreasing trend from 2012 to 2014 for Klebsiella pneumoniae (P = 0.021) and for Escherichia coli (67.5% to 58.9%), though not significant for the later one (Table 2).
Of all ESBL-screen positive bacterial strains isolated from IAIs (1900), Escherichia coli ESBL+ strains were the most frequently isolated (74.9-79.5% of all ESBL+ IAIs), followed by Klebsiella pneumoniae (16.8-23.2% of all ESBL+ IAIs) and Proteus mirabilis (1.3-3.1% of all ESBL+ IAIs) with the least frequent isolated ESBL+ strain being Klebsiella oxytoca with only 0.6-1.4% of all ESBL producing strains isolated from IAIs between 2012 and 2014. The overall percentages of ESBL production in the Escherichia coli, Klebsiella pneumoniae, Proteus mirabilis and Klebsiella oxytoca IAI strains was 36.8% with fairly constant rates for Escherichia coli, Proteus mirabilis and Klebsiella oxytoca, but a significant decrease of ESBL producing HA Klebsiella pneumoniae strains during this period (from 43.6% and 42.0% in 2012 and 2013 to 24.4% in 2014), which also reflected in a significant overall ESBL+ reduction in Escherichia coli, Klebsiella pneumoniae, Proteus mirabilis and Klebsiella oxytoca IAI isolates from 40.0% in 2012 to 37.6% in 2013 and to 32.1% in 2014 (P = 0.0038) and the overall ESBL+ production changes of Klebsiella pneumoniae within the 3 year observation period (P = 0.0215) ( Table 2). We next investigated the source of GNB strains isolated from IAIs (gall bladder, peritoneal fluid, abscess, appendix, liver and pancreas). The most frequently infected organ was the gall bladder (1072), followed by peritoneal fluid (812) and abscesses (650). The least infected organ was the pancreas (85). Escherichia coli infections occurred more frequently than Klebsiella pneumoniae infections in all organs beside the liver, in which the number of Klebsiella pneumoniae isolates (148) was higher than the Escherichia coli (88) isolates. In addition, over the three years, the highest percentages of ESBL-screen positive Escherichia coli and Klebsiella pneumoniae occurred in pancreatic isolates. In accordance with the general ESBL+ percentage decrease in Escherichia coli, isolates from the main 6 organs collected in 2014 contained less ESBL + −producing Escherichia coli strains than in 2012, which occurred for Klebsiella pneumoniae only in gall bladder, abscess, liver and appendix infections (Additional file 2: Figure S1).  (Fig. 1).

In vitro susceptibility of ESBL-screen positive
Next, we investigated local differences of susceptibilities to 12 antibiotics (Fig. 2) and the susceptibility rates against Escherichia coli and Klebsiella pneumoniae were There was no significant difference in antimicrobial susceptibility among the Escherichia coli and Klebsiella pneumoniae IAIs producing ESBL strains between HA and CA infections. However there was a tendency that susceptibility of Klebsiella pneumoniae ESBL+ infections was lower in CA than in HA IAIs (Table 3).

Discussion
The majority of IAI isolates collected in the participating centers consisted of Escherichia coli and Klebsiella pneumoniae which is similar with data from the 2002 to 2009 SMART study [16]. However, in contrast to the 2002-2009 SMART study data, which revealed an increase particularly of ESBL-screen positive Escherichia coli strains from 20.8% in 2002 up to 64.9% in 2009, in the present study ESBL+ rates in Escherichia coli strains decreased from 67.5% in 2012 to 58.9% in 2014, which was reflected also in gall bladder, abscess, liver, peritoneal fluid, appendix and pancreas derived isolates sampled in 2014. In the present study the percentages of ESBL+ Klebsiella pneumoniae strains notably dropped from 40.4% in 2012 to 26.6% in 2014 (P = 0.0215) ( Table 2), which was also seen in the decreased ESBL+ percentages of isolates from the gall bladder, abscess, liver and appendix. However, the percentages of ESBL+ Escherichia coli (66.7%) and Klebsiella pneumoniae (55.6%) strains isolated from pancreas remained high in 2014. Most of the infections occurred in the gall bladder and the peritoneum, which is in accordance with previous literature, but the number of isolates from the appendix was unusually low in our study [4,17]. In contrast to other organs, the number of Klebsiella pneumoniae liver infections exceeded those caused by Escherichia coli, which has also been reported in previous studies, and might be explained by cryptogenic infections with a new hypervirulent K1 Klebsiella pneumoniae ST23 strain, which developed in Asia and spread to Australia, European countries and the USA [18][19][20], while a recent report by Qu et al. (2015) et al. noted that K1 ST23 were the predominant Klebsiella pneumoniae liver abscess causing strains in east China [21]. ESBL-producing Escherichia coli and Klebsiella pneumoniae are supposed to be susceptible to cefoxitin, but a high proportion of these isolates tested cefoxitin-resistant (Table 3) and a likely explanation is that they have acquired AmpC beta-lactamases and porin loss, which has been described in a previous study about cefoxitin resistant Klebsiella pneumoniae strains in China, which expressed DHA-1 ß-lactamase combined with porin OmpK36 deficiency [22].
The percentages of Escherichia coli and Klebsiella pneumoniae ESBL+ strains was higher in HA than in CA IAI isolates, which is in accordance with a previous Chinese SMART study by Yang et al. (2013). However, in the latter study Escherichia coli ESBL+ rates in CA infections constantly rose from 19.1% in 2002-2003 to 61.6% in 2010-2011, whereas in our study the Escherichia coli ESBL+ rates in CA IAIs were relative constant at around 60%, with a reduction to 51.6% only in 2013. In contrast, the Escherichia coli ESBL+ rates in HA IAIs showed a decreasing trend from 70.6% in 2012 to 58.1% in 2014 in our study, but were relative stable in the years 2006-2011 (66.7%-70.0%) [23].
A more dramatic change was visible for Klebsiella pneumoniae ESBL+ rates in HA IAIs dropping from 43.6% in 2012 and 42.0% in 2013, which is similar to the Chinese HA values (39.4%) reported for Klebsiella pneumoniae IAIs in 2010-2011 [23], to 24.4% in 2014 (P = 0.006), but in CA IAIs the Klebsiella pneumoniae ESBL+ rates were relatively constant (between 24.7% and 31.0%), which is somewhat higher than the 22.2% reported for CA Klebsiella pneumoniae IAIs in 2010-2011 [23]. Taken together, the total ESBL+ rates in Klebsiella pneumoniae and Escherichia coli isolates from IAIs in our study dropped between 2012 and 2014, which was due to less ESBL+ rates in HA IAIs and rather constant percentages of CA ESBL+ Klebsiella pneumoniae and Escherichia coli IAI isolates ( Table 2). The decrease of ESBL+ GNBs might be explained by new restrictions for the clinical application of antimicrobial agents, which has been introduced by the Chinese ministry of health in 2012 [24]. The overall susceptibilities of ESBL positive Escherichia coli strains was 96.77%-98.8% to IPM, 91.26%-93.16% to ETP, 89.48%-92.75% to AMK and 84.86%-89.34% to TZP, while ESBLscreen positive Klebsiella pneumoniae strains were 70.56% -80.15% susceptible to ETP, 80.0%-87.5% to IPM, 83.82%-87.06% to AMK and 63.53%-68.38% to TZP within the three year study. However, it is noteworthy that reduced susceptibilities of Klebsiella pneumonia strains to the carbapenems IPM and ETP derived from centers located in east China indicated a local carbapenemresistance, which has also been described in other countries [25,26]. Because the eastern part of China is the most developed region with the highest incomes, antimicrobial overuse [27,28] might be an explanation for the carbapenem susceptibility difference in this area, which has been described also for the eastern Zhejiang Province before [29]. In general a previous study noted that Chinese individuals were harboring the highest number and abundance of antibiotic resistance genes in their gut microbiota compared to Danish and Spanish individuals [30], but we did not include investigations of molecular mechanisms of resistances.
All tested cephalosporins and fluoroquinolones were <70% effective for Escherichia coli and <60% for Klebsiella pneumoniae isolates that produced ESBLs. This finding is in agreement with previous literature that suggested that carbapenems are the best choice as empirical mono therapies, especially for complicated IAIs [31], but that cephalosporins, fluoroquinolones and SAM are not ideal choices for empirical treatment of IAIs in China [16]. For non-ESBL-screen positive Escherichia coli strains, cefoxitin, levofloxacin, ciprofloxacin and ampicillin-sulbactam, and for non-ESBL-screen positive Klebsiella pneumoniae strains, cefoxitin and ampicillin-sulbactam were the least effective antibiotics (Fig. 1).
Taken together, the relatively high susceptibility percentages seen for imipenem and, to a slightly lesser degree, ertapenem, against Escherichia coli -whether ESBLpositive or -negative-are important considerations in China, where ESBL+ rates around 60% are seen, and many other drugs from the beta-lactam class and fluoroquinolones are no longer viable options for therapy and should be avoided unless susceptibilities to this antimicrobial agents have been confirmed. Against Klebsiella pneumoniae, the carbapenem activity is somewhat lower, presumably because of presence of more carbapenemases (such as KPC-2-type) in Klebsiella pneumoniae [32,33], as well as other mechanisms that include porin loss combined with AmpC or ESBL enzymes; however, even the reduced activity of carbapenems to Klebsiella pneumoniae is dramatically higher than all other drugs evaluated in SMART except for amikacin and piperacillin-tazobactam. Considering the relatively high ESBL rates in China, and the low susceptibility to fluoroquinolones that are usually seen in conjunction with ESBL-positive isolates, carbapenems are among the few antimicrobial agents in China retaining sufficient in vitro activity to be considered for empiric therapy. On the other hand, it is very important to retain the activity of carbapenems, so step-down therapy to other agents should always be considered once the susceptibility of a specific pathogen is known.
A limitation of the study was that genotypic or molecular data of the strains were not included, since the SMART project does not involve these kind of analyzes.

Conclusions
From 2012 to 2014, a total of 5160 IAI isolates were obtained, of which 81.1% were caused by Enterobacteriaceae and 18.9% by non-Enterobacteriaceae, with Escherichia coli (45.4%) being the most common followed by Klebsiella pneumoniae (20.1%). The most common non-Enterobacteriaceae were Pseudomonas aeruginosa (9.8%) and Acinetobacter baumannii (6.7%). The percentages of ESBL-screen positive Escherichia coli and Klebsiella pneumoniae strains in IAI GNB isolates showed a decreasing trend from 2012 to 2014, which can be explained by less ESBL+ percentages in strains from HA IAIs.
In contrast to gall bladder, abscess, peritoneal fluid, appendix and pancreas, the percentage of Klebsiella pneumoniae causing liver infections was higher than that caused by Escherichia coli. It is noteworthy that Klebsiella pneumoniae and Escherichia coli isolates from pancreatic infections exhibited consistently high ESBL+ rates.
The apparent trend of declining percentages of ESBLscreen positive Escherichia coli and Klebsiella pneumoniae strains needs to be closely monitored.

Additional files
Additional file 1: Table S1.

Availability of data and materials
The data that support the findings of this study are directly available from MSD China Holding Co. Ltd. Data are also available from the authors upon reasonable request and with permission of MSD China Holding Co. Ltd.