Invasive liver abscess syndrome caused by Klebsiella pneumoniae

INTRODUCTION — Klebsiella pneumoniae can produce infection at a variety of sites, with the risk being increased in patients with impaired host defenses (eg, diabetes mellitus, alcoholism, malignancy, chronic obstructive pulmonary disease, and glucocorticoid therapy). K. pneumoniae is also associated with a community-acquired primary invasive liver abscess syndrome. In addition to liver abscess, some patients develop metastatic infection at other sites.

Issues related to the K. pneumoniae invasive liver abscess syndrome will be reviewed here. The epidemiology, clinical features (including the general principles of diagnosis and treatment), microbiology, and pathogenesis of K. pneumoniae infection are discussed separately. (See "Clinical features, diagnosis, and treatment of Klebsiella pneumoniae infection" and "Microbiology and pathogenesis of Klebsiella pneumoniae infection".)

DEFINITION — K. pneumoniae primary liver abscess has been variably defined in the literature. In one study, for example, primary liver abscess was defined as a K. pneumoniae liver abscess occurring in the absence of predisposing intraabdominal factors, such as hepatobiliary disease, colorectal disease, or a history of intraabdominal surgery or trauma [1]. In other studies, it has been defined by a monomicrobial K. pneumoniae isolate, while polymicrobial liver abscess was usually secondary to hepatobiliary disease or intraabdominal infection [2,3].

We prefer to define K. pneumoniae primary liver abscess (KLA) as liver abscess that occurs in the absence of hepatobiliary disease. Almost all of these infections are monomicrobial [3-5]. Although this topic highlights virulent strains of K. pneumoniae that have the capacity to invade a healthy liver without predisposing anatomical abnormalities, the rare possibility of concurrent colorectal disease cannot be excluded, as routine colonoscopy is not generally performed in the setting of KLA. The risk of colorectal cancer in the setting of pyogenic liver abscess is discussed elsewhere. (See "Pyogenic liver abscess", section on 'Association with colorectal neoplasia'.)

EPIDEMIOLOGY — Most cases of K. pneumoniae primary liver abscess (KLA), particularly those associated with metastatic infection, have been reported in Taiwan and are community-acquired [1-12]. In a series of 248 patients with pyogenic liver abscess from Taiwan, for example, K. pneumoniae was responsible for 171 (69 percent) [4].

Community-acquired KLA has also been described in other countries in Asia [7,13-17], in Asian patients living in other countries [18-24], and in South Africa [25]. It has less commonly been reported in non-Asian patients in the United States, Europe, Canada, and South America [19,20,26-38]. A report from New York, for example, evaluated 79 cases of liver abscess [19]. K. pneumoniae was the most commonly identified pathogen, isolated from 23 of 54 liver abscesses (41 percent) in which an organism was recovered. K. pneumoniae was more commonly isolated among Asian than non-Asian patients (50 versus 27 percent).

A minority of patients with KLA, mostly from Taiwan, develops metastatic infection, most commonly manifested as endophthalmitis and/or meningitis [1,6,8,10,39,40]. The incidence of metastatic infection in the setting of KLA is about 12 percent in series from Taiwan and is more common in KLA than in liver abscesses of other bacterial etiology (15 versus 4 percent) [1,3,4]. A somewhat lower rate of metastatic infection (9 percent) was noted in a series of 290 patients with KLA from Korea [13]. (See 'Metastatic infection' below.)

The findings have been more variable in studies in the United States. There were no cases of metastatic infection among the 23 patients with KLA in a study from New York [19], whereas a review of 18 cases in the United States prior to the New York study found metastatic infection in five (28 percent) [20].

In contrast, metastatic infection appears to be a rare complication (2 percent or less) of secondary K. pneumoniae liver abscess, even in Taiwan [1,3].

Metastatic infection can also occur with K. pneumoniae infections other than liver abscess. These infections are discussed separately. (See "Clinical features, diagnosis, and treatment of Klebsiella pneumoniae infection", section on 'Endophthalmitis' and "Clinical features, diagnosis, and treatment of Klebsiella pneumoniae infection", section on 'Meningitis/brain abscess'.)

PATHOGENESIS AND RISK FACTORS — Klebsiella primary liver abscess (KLA) frequently occurs in patients with diabetes but can occur in the absence of underlying predisposing medical conditions.

Host factors

Diabetes mellitus — Diabetes mellitus or impaired fasting glucose is the major observed risk factor for KLA; these conditions are not major risk factors for liver abscesses caused by organisms other than Klebsiella [3-6,12,13].

The mechanism by which diabetes predisposes to KLA is not well understood. One contributory mechanism may be that poor glycemic control impairs neutrophil phagocytosis of K1 and K2 capsular serotypes, although diabetic patients are also vulnerable to infection caused by non-K1/K2 klebsiella strains [41-43]. (See "Microbiology and pathogenesis of Klebsiella pneumoniae infection", section on 'Association with primary liver abscess'.)

Other host factors — Fatty liver disease has also been more highly associated with KLA compared with liver abscesses due to other organisms [44].

Although no studies have identified specific human genes that predispose to KLA, the prevalence in Asian patients has raised the possibility that host genetic factors play a role. As an example, Chinese ethnicity itself might be a major factor predisposing to intestinal colonization by serotype K1/K2 K. pneumoniae isolates, which is associated with KLA. In a seroprevalence study that evaluated K. pneumoniae isolates from the stools of healthy adult Chinese residents of Taiwan, Japan, Hong Kong, China, Thailand, Malaysia, Singapore, and Vietnam, 10 percent of the isolates were serotypes K1/K2 [45]. However, K1/K2 isolates in stool do not always correlate with the microbiology of KLA isolates; in a study of 11 patients in Korea with K1/K2 KLA and Klebsiella spp in their stool, only 27 percent had concordant species in liver and stool [46]. (See "Microbiology and pathogenesis of Klebsiella pneumoniae infection", section on 'Pathogenesis' and "Microbiology and pathogenesis of Klebsiella pneumoniae infection", section on 'Association with primary liver abscess'.)

Virulence factors — KLA is caused by isolates with an increased number of virulence factors compared with other K. pneumoniae isolates [1-3,6,11,25].

In studies from Taiwan, both the K1 capsular serotype and the hypermucoviscosity phenotype are more common in community-acquired compared with nosocomial K. pneumoniae infections, including KLA [1-3,6,11,25]. In a report from Taiwan of 95 K. pneumoniae isolates from secondary bacteremia with an identified source of infection, abscess formation in the liver and other sites was much more common with community-acquired than nosocomial isolates (46 versus 4 percent) [2].

The regional distribution of K. pneumoniae virulence factors also appears to play a significant role in the geographic restriction of KLA. In a review of 455 consecutive cases of K. pneumoniae bacteremia in seven countries (Taiwan, South Africa, United States, Australia, Belgium, Turkey, and Argentina), the invasive community-acquired syndrome of liver abscess, meningitis, or endophthalmitis was only seen in Taiwan and South Africa [25]. The isolates from Taiwan and South Africa compared with the other countries were much more likely to have a mucoid phenotype (100 versus 2 percent) and to be rmpA-positive (86 versus 7 percent). Other virulence factors, such as K1 capsular serotype, were common in Taiwan and South Africa but rarely seen in the other countries.

The role of virulence factors in KLA is discussed in further detail separately. (See "Microbiology and pathogenesis of Klebsiella pneumoniae infection", section on 'Pathogenesis' and "Microbiology and pathogenesis of Klebsiella pneumoniae infection", section on 'Association with primary liver abscess'.)

Antibiotic use — Prior antibiotic use also appears to increase the risk of KLA. As an example, in a retrospective study of 855 patients with KLA and 3400 age- and sex-matched controls in Taiwan, ampicillin or amoxicillin therapy within the prior 30 days was associated with KLA [47]. In an accompanying animal study, ampicillin administration predisposed K. pneumoniae-colonized mice to increased liver abscess formation.

Risk factors for metastatic disease — Existing published data are conflicting as to whether diabetes is an independent risk factor for metastatic infection [1,3,6,12,48,49]. Virulence factors of the isolate have been more clearly associated with the development of metastatic disease. The presence of the more virulent K1 serotype is a risk factor for metastatic infection [1,6]. In a report from Taiwan of patients with primary Klebsiella liver abscess (KLA), septic endophthalmitis was present in infections with 12 of 85 K1 strains, 2 of 19 K2 strains, and none of 28 non-K1/K2 strains [6]. A similar relationship was seen in the four patients with other sites of metastatic infection: none occurred in the non-K1/K2 strains. (See "Microbiology and pathogenesis of Klebsiella pneumoniae infection", section on 'Capsular serotypes'.)

Presence of the magA gene is another risk factor for metastatic infection [8]. Among magA-positive isolates, those that also have the kfu iron-uptake system may be particularly associated with metastatic infection [50]. The rmpA gene also appears to be a significant predictor of metastatic infection [51]. However, the rmpA gene is almost ubiquitous in KLA strains, thus rmpA could hardly predict metastatic infections among patients with KLA [52]. (See "Microbiology and pathogenesis of Klebsiella pneumoniae infection", section on 'Siderophores'.)

CLINICAL MANIFESTATIONS — The presenting symptoms and signs of K. pneumoniae liver abscess (KLA) are similar to those caused by other etiologies of pyogenic liver abscess [3,4,53]. (See "Pyogenic liver abscess", section on 'Clinical manifestations'.)

In a review of 160 cases from Taiwan, the most common clinical features and their frequencies are listed below [3]:

●Fever (93 percent)

●Right upper quadrant tenderness (71 percent)

●Nausea, vomiting, diarrhea, or abdominal pain (38 percent)

●Leukocytosis (70 percent)

●Elevations in serum alanine and aspartate aminotransferases (59 and 68 percent)

●Elevations in alkaline phosphatase (78 percent)

●Elevations in bilirubin (26 percent)

In contrast to liver abscesses caused by other organisms, those due to K. pneumoniae are more likely to be solitary and more likely to be monomicrobial [3-5]. In a report from Taiwan, for example, 65 of 68 were monomicrobial [5].

Certain imaging findings have been reported more commonly with KLA compared with liver abscesses of other bacterial etiology. On ultrasonography, KLA often has a predominantly solid appearance and, compared with other causes of bacterial liver abscess, has a much smaller quantity of pus at initial aspiration [54]. On computed tomography (CT) scan, monomicrobial KLA tends to be a single, multiloculated abscess involving only one lobe of the liver [55]. When compared with abscesses caused by other bacteria, KLA is more likely to be solid, thin-walled, without rim enhancement, and associated with septic thrombophlebitis of the portal or hepatic venous systems [55-57]. In one study of 169 patients with KLA, regional thrombophlebitis of the portal and/or hepatic veins was identified in 31 percent and improved following abscess resolution [58].

Metastatic infection — A minority of patients with primary liver abscess have concurrent evidence of or develops metastatic infection at other sites [1,3,4,6,8,10,12,13,20,39,40,50,59-61]. Thrombophlebitis of the portal or hepatic venous systems could be a predisposing condition to metastatic disease [62]. The most common manifestations of metastatic infection are endophthalmitis, meningitis and brain abscess [1,3,12]. Other manifestations include lumbar or cervical spondylitis and discitis, septic pulmonary emboli, lung abscess, psoas abscess, splenic abscess, necrotizing fasciitis, neck abscess, and osteomyelitis [1,3,12,18,63].

In one study of 66 patients with KLA in China, metastatic infection occurred in 22 (33 percent); of those, septic emboli occurred in eight patients, extrapulmonary metastatic infection occurred in six, and both pulmonary and extrapulmonary disease occurred in eight [64]. Septic pulmonary emboli were associated with a unilocular abscess whereas extrapulmonary metastatic infection was associated with a smaller diameter (<5 mm) abscess.

Transdiaphragmatic extension and perforation into the pericardium causing purulent pericarditis, cardiac tamponade, and/or pleural empyema have been rarely reported [65,66] (See "Clinical features, diagnosis, and treatment of Klebsiella pneumoniae infection", section on 'Endophthalmitis' and "Clinical features, diagnosis, and treatment of Klebsiella pneumoniae infection", section on 'Meningitis/brain abscess'.)

DIAGNOSIS — Similar to the diagnosis of pyogenic liver abscesses, the diagnosis of primary liver abscess due to K. pneumoniae (KLA) is made when K. pneumoniae is isolated from the abscess aspirate or blood of a patient with imaging findings consistent with a liver abscess in the absence of underlying hepatobiliary disease. (See "Pyogenic liver abscess", section on 'Diagnosis'.)

Ultrasonography and computed tomography (CT) are the imaging methods of choice when liver abscess is part of the differential diagnosis in patients with the above clinical manifestations and laboratory abnormalities (image 1). Imaging for liver abscess may also be warranted in patients who present with K. pneumoniae bacteremia and have persistent fever despite appropriate antibiotic therapy. Some radiographic findings have been associated with KLA compared with liver abscesses due to other pathogens, but these findings alone are not diagnostic of KLA. (See 'Clinical manifestations' above.)

Following radiologic identification of a possible liver abscess, image-guided diagnostic aspiration should be performed to confirm a pyogenic process. Specimens should be sent for Gram stain and both aerobic and anaerobic cultures. Blood cultures should be obtained from all patients with known or suspected liver abscess since KLA may be associated with bacteremia [10,11,25].

DIFFERENTIAL DIAGNOSIS — The differential diagnosis of fever and right upper quadrant abdominal pain includes other hepatobiliary disease, colitis, or pneumonia. This topic is reviewed in detail separately.

The differential diagnosis of patients presenting with a focal liver lesion is broad and includes both malignant and infectious etiologies. Imaging studies (ultrasound or computed tomography [CT]) can usually differentiate abscess from malignancy. In some instances, however, a multiloculated abscess may mimic the heterogeneity of hepatic tumor (image 2). Conversely, necrosis of hepatic tumor can appear similar to a highly suppurative abscess on CT. In these diagnostic difficulties, diffusion-weighted magnetic resonance imaging may be useful to differentiate between hepatic abscess and necrotic liver tumor [67,68]. However, it is possible that K. pneumoniae could be isolated from an infected and necrotic hepatic tumor [69,70]. In such a situation, the diagnosis is extremely difficult and may require tissue biopsies. This is a particular consideration in the setting of a non-resolving liver abscess. (See "Approach to the adult patient with an incidental solid liver lesion".)

In a patient with imaging findings consistent with a liver abscess, many different microbiological etiologies are possible and cannot be distinguished without culture or serological data. (See "Pyogenic liver abscess", section on 'Microbiology'.)

TREATMENT — Treatment of primary Klebsiella liver abscess (KLA) involves drainage and systemic antibiotic therapy.

Drainage — Percutaneous drainage guided by imaging (either ultrasonography or computed tomography [CT]) is used for both diagnosis and treatment and is preferred over surgical drainage. Percutaneous drainage for treatment is recommended even in patients in whom the diagnosis has been made on the basis of positive blood cultures.

In some cases, the abscess may not be amenable to immediate drainage, for example, if it is still in an immature form. Two options are to postpone drainage and monitor closely until the abscess has matured and can then be drained or to insert the drain and retain the tubing so that the abscess can drain once it becomes liquefied. Surgical resection may be warranted in some instances, such as when the abscess is multiloculated, it fails to liquefy, or there is delayed resolution and a protracted course of fever. (See "Pyogenic liver abscess", section on 'Drainage'.)

Antibiotic choice — Treatment of KLA requires parenteral antibiotic therapy in addition to drainage. Generally, community-acquired KLA isolates remain susceptible to cephalosporins [14,71]. Among pyogenic liver abscesses caused by multidrug-resistant K. pneumoniae, most occur in patients with other hepatobiliary disease (as opposed to primary KLA) [72]. Nevertheless, resistant isolates, including those that produce extended-spectrum beta-lactamases (ESBLs), have been rarely reported to cause KLA [73-76]. Furthermore, KLA caused by unusual carbapenem-resistant virulent strains has been emerging in China [77]. Initial therapy for KLA can be administered as outlined for empiric management of pyogenic liver abscess, which generally should include coverage for gram-negative and anaerobic organisms (table 1). (See "Pyogenic liver abscess", section on 'Antibiotic therapy'.)

Subsequent antibiotic therapy should be tailored to results of antibiotic susceptibility testing. In reports from Taiwan and Korea, antibiotic regimens have included extended spectrum beta-lactams and cephalosporins with or without aminoglycosides [1,3,14,20].

There is disagreement regarding the selection of early or late generation cephalosporins. Many favor use of first generation cephalosporins (with or without an aminoglycoside) given their relatively low cost and apparent efficacy with respect to rates of mortality, metastatic infection, and complications [51]. However, others have reported higher metastatic infection rates among patients treated with cefazolin compared with those treated with a second or third generation cephalosporin, both with or without an aminoglycoside (37 versus 6 percent) [78].

In regions where third generation cephalosporins are much more costly than first generation cephalosporins, cefazolin with or without an aminoglycoside may be the favored antibiotic regimen. However, third generation cephalosporins are favorable if costs are not prohibitive and/or in the setting of severe infection.

Although aminoglycosides penetrate abscess cavities poorly, in theory they may eradicate bloodstream organisms early in the course of infection, potentially decreasing risk for metastatic complications. However, this benefit is unproven and may be outweighed by the toxicity of aminoglycosides. (See "Aminoglycosides".)

Duration of therapy — For most cases, antibiotic therapy should be administered for four to six weeks. Longer courses of treatment may be warranted for patients requiring subsequent drainage procedures or with persistent radiographic evidence of abscess. Parenteral antibiotics may be administered for the first two to three weeks until the patient has improved systemically and drainage is complete; the remainder of the course can be completed with oral agents.

Follow-up imaging should be used to monitor the response to therapy, to determine the duration of antibiotics, and to assess the need for further aspiration. In general, treatment should be continued until CT imaging demonstrates complete or near complete resolution of the abscess cavity. Some abscesses with a thick capsule may leave a CT abnormality remaining after the infection has been eradicated, and other clinical findings, such as normalization of inflammatory markers, sterile cultures, and the finding of a clear cavity without debris or other content on imaging, can be helpful to distinguish this from persistent infection.

Treatment of metastatic infection — In addition to systemic antibiotics, local therapy or debridement may be required in patients with metastatic infection. As an example, patients with Klebsiella endophthalmitis should receive intravitreal antibiotics and vitrectomy. Management of the metastatic complications is discussed further in the topic reviews on those clinical syndromes. (See "Bacterial endophthalmitis".)

PREVENTION — Data are scarce regarding whether certain interventions can prevent KLA. A reasonable approach to mitigate known risk factors includes preventing or controlling diabetes mellitus, fatty liver disease, and unnecessary antibiotic usage. Observational data suggest that patients who take certain medications may have lower risk of KLA; such medications include aspirin, clopidogrel, metformin, H2 blockers, and proton pump inhibitors [79,80].

PROGNOSIS — The prognosis of K. pneumoniae liver abscess (KLA) is good overall, but the metastatic complications can cause significant morbidity. The largest published experience comes from Taiwan:

●In a review of 160 cases of KLA (all but one were primary), the mortality rate was 11 percent and relapse occurred in 4.4 percent [3]. Among the 18 patients who died, the main causes of death were fulminant sepsis in nine (all with inadequate or delayed drainage of the abscess), metastatic infection in four (meningitis in two, brain and lung abscess in one, and necrotizing fasciitis in one), and rupture of the abscess in two. The mortality rate was significantly higher (41 percent) in 22 patients with polymicrobial liver abscess, all but one of whom had a biliary tract stone or intraabdominal malignancy. Sepsis was the only cause of death in these patients.

●A later study evaluated 248 patients with pyogenic liver abscess (171 due to K. pneumoniae and 77 to other bacteria) [4]. The mortality rate was significantly lower in patients with K. pneumoniae infection (4.1 versus 20.8 percent), while the relapse rate was the same in the two groups (6.5 versus 6.4 percent).

●In a study of 110 patients with a first episode of KLA in China, the recurrence rate was 18 percent, and extended-spectrum beta-lactamase (ESBL) production was independently associated with recurrence [81].

Septic pulmonary embolism caused by a KLA, combined with extrapulmonary metastatic infection of other vital organs like meningitis or pericarditis, confers a poor prognosis [82].

Although the mortality rate is relatively low, the morbidity in patients with metastatic endophthalmitis is often high despite aggressive therapy, as many patients have impaired vision or blindness [1,61,83-86]. In addition, patients with meningitis may have persistent neurologic abnormalities [1,87], particularly if they have substantial neurologic impairment prior to the onset of antibiotic therapy [87]. Among patients with septic pulmonary emboli in the setting of KLA, other extrapulmonary metastatic infection is associated with a poor prognosis [82].

The frequency of adverse outcomes was illustrated in a report of 23 patients with KLA who had metastatic infection involving the eye or central nervous system [1]. Sixteen had severe irreversible disability, including loss of vision, quadriplegia, paraparesis, or impaired higher cortical function. Good vision at presentation and early therapy have been associated with a higher likelihood of maintenance of vision [83,84]. (See "Clinical features, diagnosis, and treatment of Klebsiella pneumoniae infection", section on 'Endophthalmitis' and "Clinical features, diagnosis, and treatment of Klebsiella pneumoniae infection", section on 'Meningitis/brain abscess'.)

The higher mortality rates with hypervirulent strains may reflect the impact of comorbidities rather than virulence factors [88].

SUMMARY AND RECOMMENDATIONS

●Epidemiology – Klebsiella pneumoniae primary liver abscess (KLA) occurs in the absence of hepatobiliary disease and is almost always monomicrobial. Most cases have been reported from Asia or in patients of Asian origin. (See 'Definition' above and 'Epidemiology' above.)

●Risk factors – Diabetes mellitus or impaired fasting glucose is the most important host risk factor for primary KLA. (See 'Host factors' above.)

●Virulence factors – The Klebsiella isolates that cause KLA have an increased number of virulence factors compared with other Klebsiella isolates and appear to be restricted geographically. (See 'Virulence factors' above.)

●Clinical manifestations – In addition to the manifestations typical of pyogenic liver abscess, such as fever, leukocytosis, right upper quadrant tenderness, and elevated liver enzymes, a minority of patients with primary KLA can develop metastatic infections at other sites. The most common sites for metastatic infections are the eye, meninges, and brain. (See 'Clinical manifestations' above.)

●Diagnosis – Imaging should be performed in patients with signs and symptoms of a liver abscess or in patients with K. pneumoniae bacteremia who have persistent fevers despite appropriate antibiotic therapy. Diagnosis of primary KLA is made by detection of a liver abscess on imaging (ultrasound or computed tomography [CT]) followed by aspiration of the lesion for Gram stain and aerobic and anaerobic culture. (See 'Diagnosis' above.)

●Management – Treatment of KLA requires parenteral antibiotic therapy in addition to drainage, optimally percutaneous. Antibiotic choice should be based upon the results of antibiotic susceptibility testing. A third-generation cephalosporin is preferable if the isolate is susceptible and cost is not prohibitive. Antibiotics should be given for at least four to six weeks, depending on abscess resolution as determined by imaging findings. (See 'Antibiotic choice' above and 'Duration of therapy' above.)

●Prognosis – The reported mortality rate has ranged from 4 to 11 percent. Metastatic disease to the eyes or brain can cause significant long-term morbidity. (See 'Prognosis' above.)

ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges Yin-Ching Chuang, MD, who contributed to earlier versions of this topic review.