The most commonly identified community-associated infections caused by K. pneumoniae include urinary tract infections (UTIs), pneumonia, and liver abscesses. Community-acquired liver abscesses caused by hypervirulent strains (for example, the K1 capsular serotype) of K. pneumoniae have become increasingly common among persons without underlying hepatobiliary disease, particularly in Asia’s Pacific Rim. One potential explanation for these geographic differences in the incidence of disease is a higher rate of intestinal carriage of K. pneumoniae of the K1 serotype among persons living in this region. In the previously mentioned Korean study, among the 21.1% of healthy adults who had fecal carriage of K. pneumoniae, 23% were carriers of serotype K1 ² . Interestingly, the rates of carriage of the K1 serotype were 5.6% among adults older than 25 years of age and 0% among those 16–25 years of age ( P=0.007), and the prevalence of the K1 serotype was higher among isolates from ethnic Koreans residing in Korea than among those from persons residing in other countries (24.1% versus 5.6%). These findings suggest that environmental factors, such as diet, may play a role in the epidemiologic differences observed with regard to this pathogen ² .

In the laboratory, these hypervirulent strains are phenotypically characterized by hypermucoviscosity often defined by a positive “string test” in which a viscous string of more than 5 mm in length forms when bacterial colonies on an agar plate are stretched by an inoculation loop. Acquisition of an rmpA gene-containing plasmid has been associated with this phenotype through rmpA-mediated increases in capsule production. Explanations for the increased virulence of these strains include resistance to complement and neutrophil-mediated bactericidal activity ³ . Liver abscesses caused by these particular serotypes are associated with an increased risk of hematogenous dissemination to other sites (for example, lung, eye, and central nervous system) and with relatively high rates of morbidity and mortality (4–8%). Evaluation for hepatic abscess should be considered in persons presenting with K. pneumoniae bloodstream infection, endophthalmitis, or central nervous system infection without another explanation, particularly among persons with epidemiologic risk factors (for example, residence in or travel to endemic regions of Asia). Extended-spectrum beta-lactamases (ESBLs) and AmpC beta-lactamases are fortunately not common among these strains but have been reported ^{4, 5} .

Although K. pneumoniae is a relatively infrequent cause of community-acquired infections, it is a common cause of healthcare-associated infections. In fact, Klebsiella species (that is, K. pneumoniae and K. oxytoca) were the third most common pathogens reported among all cases of central line-associated bloodstream infection (CLABSI), catheter-associated urinary tract infection (CAUTI), ventilator-associated pneumonia, and surgical site infection reported to the National Healthcare Safety Network (NHSN) of the Centers for Disease Control and Prevention (CDC) between 2011 and 2014, representing 7.7% of all reported pathogens ⁶ . Over the past few years, several outbreaks of infection caused by multidrug-resistant strains of K. pneumoniae and other Enterobacteriaceae associated with the use of contaminated duodenoscopes have been reported. These outbreaks have been attributed at least in part to the complex design of duodenoscopes that renders them more difficult to adequately clean and disinfect as compared with endoscopes with simpler designs ^{7– 10} . Although these duodenoscope-related infections have largely been associated with carbapenem-resistant K. pneumoniae (CRKP) and other carbapenem-resistant Enterobacteriaceae (CRE), these infections may represent just the tip of the iceberg of duodenoscope-related infections as infections caused by organisms with less notable resistance patterns may be less likely to trigger further investigation of a possible device-associated infection.

The increased incidence of K. pneumoniae infection among persons with current or recent healthcare exposures appears to be due at least in part to increased rates and/or bacterial burdens of gastrointestinal carriage of the organism, which has been identified as a risk factor for subsequent K. pneumoniae infection. For example, in a prospective cohort study of 498 patients, rates of carriage of K. pneumoniae, as detected by rectal and throat samples collected shortly after admission to the intensive care unit (ICU), were approximately 6% among patients admitted directly from the community and 19% among patients with recent healthcare contacts ¹¹ . In that study, K. pneumoniae colonization was found to be a significant risk factor for subsequent K. pneumoniae infection while in the ICU, and 16% of carriers versus 3% of non-carriers developed infection and approximately half of K. pneumoniae infections came from the patients’ own microbiota. Similarly, another study of patients hospitalized in ICUs and oncology wards found a prevalence of rectal carriage of K. pneumoniae of 23% ¹² . Among those identified as carriers, 5.2% subsequently developed K. pneumoniae infection (versus 1.3% of non-colonized patients, odds ratio (OR) 4.06, P<0.001). Finally, 27% of ICU patients identified as asymptomatic carriers of CRKP through perianal swab specimens collected as part of an active surveillance program subsequently developed CRKP bloodstream infection during their index hospitalization ¹³ .

In a 2013 report, the CDC identified CRE, the most common of which is K. pneumoniae, as one of the top three urgent antimicrobial resistance threats ¹⁶ . This report estimated that, in the US each year, there are 7,900 infections and 520 deaths due to carbapenem-resistant Klebsiella species. KPC is the most common mechanism of carbapenem resistance among K. pneumoniae isolates in the US. As of January 2017, KPC-producing CRE had been reported to the CDC from 48 states in the US ¹⁷ . KPC-producing CRE typically also have genes that confer resistance to several other classes of antimicrobials, resulting in these organisms displaying a multidrug-resistant phenotype. Among CRE isolates reported to the CDC as of January 6, 2017, carbapenemases other than KPC that have been identified include NDM (n=175), oxacillinase-48-type carbapenemase (OXA-48, n=73), Verona integron-encoded metallo-beta-lactamase (VIM, n=27), and imipenemase metallo-beta-lactamase (IMP, n=11). There are other mechanisms of carbapenem resistance, including hyperproduction of the AmpC beta-lactamase with coexisting porin mutations. In some regions outside the US, non-KPC mechanisms of resistance predominate.

The polymyxins—polymyxin E (or colistin) and polymyxin B—serve as drugs of last resort for treatment of infections caused by CRKP and other multidrug-resistant gram-negative pathogens; however, resistance to the polymyxins is a well-described phenomenon and has become increasingly common in some areas, and a rate of 27% was reported among KPC-producing K. pneumoniae isolates from five Italian hospitals in 2013 ¹⁸ . Between 2009 and 2012, colistin resistance was detected in 2.7% of Klebsiella isolates obtained from patients hospitalized with pneumonia at one of 28 medical centers in the US ¹⁹ . Polymyxin resistance has been reported as an independent predictor of 14-day mortality among patients with KPC-producing K. pneumoniae infections ¹⁸ .

Resistance to polymyxin is most commonly chromosomally mediated and this is typically due to post-translational modification of bacterial outer-membrane lipopolysaccharide through the addition of cationic groups as a result of mutations in the mgrB regulatory gene ^{20– 24} . Previous treatment with colistin has been associated with infection due to colistin-resistant KPC-producing K. pneumoniae ²⁵ ; however, the majority of patients with colistin-resistant KPC- K. pneumoniae bloodstream infection in one study had not previously received colistin, indicating that other factors, such as healthcare-associated transmission of resistant strains, likely also play a role. More recently, a plasmid-mediated gene ( mcr-1) that confers resistance to polymyxin was discovered. The gene has been identified among Enterobacteriaceae (most commonly E. coli but also other genera, including Klebsiella) isolated from animal and human sources. The initial report from China in November 2015 includes isolates obtained as early as 2011 ²⁶ . In that investigation, the rates of prevalence of the mcr-1 gene were 20.6% among E. coli isolates obtained from pigs at the time of slaughter, 14.9% among E. coli isolates obtained from retail chicken and pork, and 1.4% and 0.7% among E. coli and K. pneumoniae isolates, respectively, from human inpatients. Since that report, mcr-1 has been detected in food animals, food, or humans (or a combination of these) in other Asian countries, Europe, Africa, North America, and South America ^{27, 28} . As of March 30, 2017, 14 mcr-1-positive isolates, including 12 human and two animal isolates, had been reported from 11 US states ²⁹ . Investigators have demonstrated successful in vitro transfer of the mcr-1 gene between E. coli isolates by conjugation and from E. coli to E. coli, K. pneumoniae, and Pseudomonas aeruginosa by transformation ²⁶ , raising significant concerns about the possibility of acquisition of this mobile genetic element by organisms that already possess other drug-resistance determinants (for example, KPC) resulting in pandrug-resistant bacteria.

Accurate assessment of polymyxin susceptibility among multidrug-resistant strains of K. pneumoniae is important for both clinical and epidemiologic purposes. Polymyxin susceptibility testing is challenging for several reasons (for example, poor diffusion of polymyxins into agar and adherence of polymyxins to components of some materials used in the laboratory) and these challenges can lead to unacceptably high error rates when methods such as disk and agar gradient diffusion are used. The Clinical Laboratory Standards Institute/European Committee on Antimicrobial Susceptibility Testing (CLSI/EUCAST) Joint Polymyxin Breakpoints Working Group recently identified broth microdilution as the reference method for testing colistin susceptibility ³⁰ . No specific statements or recommendations for polymyxin B susceptibility testing were made, but many of the same limitations and challenges as noted for colistin are applicable to polymyxin B as well.

Some but not all retrospective studies have found an association between combination therapy and lower mortality as compared with treatment with single-drug treatment regimens. One of the more common combination therapy regimens is that of a polymyxin plus an optimized dosing regimen of a carbapenem (for example, meropenem or doripenem). Some observational studies have reported better outcomes among patients who received polymyxin-carbapenem combination therapy, particularly in patients or settings (or both) in which K. pneumoniae isolates demonstrated relatively low-level carbapenem resistance (MIC ≤8–16) ^{18, 34} . Retrospective studies that have assessed this strategy among patients infected with CRKP with high-level carbapenem resistance have generally not demonstrated additional benefit ^{35, 36} . These MIC-dependent differences in clinical outcomes are supported by in vitro testing that has demonstrated synergy between bactericidal activity of polymyxin B plus meropenem against KPC-2-producing K. pneumoniae isolates with meropenem MICs of not more than 16 μg/mL ³⁷ . Similarly, exposure to this combination of agents produced morphological changes in bacterial isolates that decreased as the meropenem MIC increased. The pharmacodynamic effects of combination therapy decreased as the meropenem MIC increased to more than 16 μg/mL. Thus, combinations of a polymyxin plus optimized dosing of a carbapenem appear to be most likely of benefit for treatment of infections caused by CRKP strains with relatively low-level carbapenem resistance (that is, MIC ≤8–16 μg/mL).

Dual carbapenem therapy refers to the use of ertapenem, which is the carbapenem with the highest affinity for carbapenemase enzymes, in addition to a carbapenem that has lower affinity for the enzyme (for example, meropenem or doripenem) so that ertapenem preferentially consumes the carbapenemase, allowing a larger proportion of the other carbapenem to remain intact and to exert its antimicrobial effect on the infecting organism. In vitro or animal models using both immunocompetent and neutropenic mice (or both) have demonstrated enhanced efficacy of dual carbapenem regimens (for example, optimized-dose doripenem plus ertapenem) as compared with a single carbapenem (doripenem) against at least some isolates of KPC- and NDM-producing K. pneumoniae ^{38, 39} . This strategy appears to be most likely effective for treatment of infections caused by organisms with relatively low carbapenem MICs (for example, ≤4–16 µg/mL) and may be less effective in immunocompromised hosts ⁴⁰ . Human clinical data are limited, but at least two small case series of three patients each with high-level CRKP bacteremia or UTI have reported successful treatment using the combination of ertapenem plus either doripenem or meropenem ^{41, 42} .

The two previously described treatment strategies both involve the use of optimized dosing of a carbapenem (for example, meropenem or doripenem). This refers to administering the drug in doses and frequencies that are most likely to allow the free drug concentration to exceed the MIC of the infecting organism ( fT>MIC) for at least 40% of the dosing interval. Treatment regimens using higher doses and prolonged infusion of carbapenems (for example, meropenem 2 grams intravenously every 8 hours infused over the course of 3 hours) have achieved this fT>MIC goal and have been successful in the treatment of infections due to other gram-negative organisms with relatively low-level carbapenem resistance MIC (<16 µg/mL), including P. aeruginosa. However, at least one in vitro pharmacodynamic model was not reliably able to reach a fT>MIC of at least 40% for KPC-producing K. pneumoniae isolates with MICs of 8 µg/mL and did not achieve a fT>MIC of at least 40% for any isolates with an MIC of 16 µg/mL due to rapid hydrolysis of meropenem ⁴³ .

Ceftazidime-avibactam was approved by the US Food and Drug Administration (FDA) in 2015 for treatment of complicated UTI, including pyelonephritis and, in combination with metronidazole, complicated intra-abdominal infections. Avibactam is a new non-beta-lactam beta-lactamase inhibitor that has activity against class A and C beta-lactamases, including ESBLs, AmpC, and KPC. It is not active against class B metallo-beta-lactamases, such as NDM ⁴⁴ . Testing of collections of clinical isolates suggests that ceftazidime-avibactam has activity against most isolates of multidrug-resistant K. pneumoniae isolates (FDA breakpoint MIC value ≤8 µg/mL) ^{45– 47} . There are currently no data from randomized controlled trials of the use of ceftazidime-avibactam for the treatment of CRKP, but available data from case series suggest that ceftazidime-avibactam may offer a useful therapeutic option for some patients with CRKP infections ^{48, 49} . Temkin and colleagues described the use of ceftazidime-avibactam for the treatment of 36 patients with CRE infection (34 of which were caused by CRKP) and two patients with carbapenem-resistant P. aeruginosa ⁴⁸ . Among those 38 patients, 28 (74%) experienced clinical or microbiological cure or both. Shields and colleagues described their experience using ceftazidime-avibactam to treat 37 patients with CRE infection, 84% of which were K. pneumoniae with 90% of those harboring a bla _KPC gene ⁴⁹ . Clinical success was achieved in 59% of the patients. Microbiologic failure occurred in 27% of the patients and development of ceftazidime-avibactam resistance (that is, MIC >8 µg/mL) was detected in three 3(30%) of 10 cases within 10–19 days of ceftazidime-avibactam exposure. In subsequent investigations, emergence of mutations in bla _KPC-3 was detected among the three isolates that had acquired ceftazidime-avibactam resistance. The resulting variant enzymes were demonstrated to result in resistance to ceftazidime-avibactam; however, these variant enzymes had reduced carbapenamase activity ⁵⁰ . Meropenem MICs were reduced by at least 4-fold from baseline with restoration of meropenem susceptibility in two of the three isolates. Other identified mechanisms of resistance to ceftazidime-avibactam among KPC-producing Enterobacteriaceae include production of large amounts of AmpC beta-lactamase ⁵¹ , suggesting that ceftazidime-avibactam may not be active against organisms with this particular combination of beta-lactamases, and mutations in outer-membrane porins (for example, ompK35) ^{35, 52} .

Until very recently, the pipeline of new drugs for the treatment of multidrug-resistant gram-negative pathogens had been described as “dry”. However, a number of new drugs with in vitro activity against multidrug-resistant strains of K. pneumoniae are now in development. These include combinations of existing beta-lactam drugs with new carbapenemase inhibitors and novel drugs within existing drug classes (for example, cephalosporins and aminoglycosides). Table 1 describes drugs that are currently in phase 3 clinical trials or under FDA review or both. Additional investigational drugs are currently being evaluated in phase 1 and phase 2 studies. The development of novel therapeutic agents with activity against multidrug-resistant strains of K. pneumoniae and other gram-negative organisms is urgently needed.