In Asia, it has been confirmed that Klebsiella pneumoniae (KP) has replaced other pathogens as the primary pathogenic bacterium of community-acquired pyogenic liver abscess. Particularly in Taiwan, mainland China, South Korea, and India, the detection rates of KP soar to between 70% and 90% (Liu et al., 1986; Yin et al., 2021; Behera et al., 2025; Dhamelia et al., 2025; Juan et al., 2025; Si et al., 2025). The elevated incidence is tightly linked to the pervasive presence of K1/K2 serotypes in hypervirulent KP (hvKP) strains and their associated virulence genes (Ko et al., 2002; Chung et al., 2007). Serotypes K1 and K2 are predominant. A study conducted in Shandong, China has found that the detection rates for K1 and K2 in hvKP strains are 25.00% and 36.67%, respectively, which are significantly higher than the detection rates in classical KP strains, at 12.00% and 9.20% (Han et al., 2025). These patterns are also noted in other Asian countries such as Iran and Malaysia, where K1 and K2 serotypes frequently coexist with hvKP virulence genes, thereby amplifying their pathogenic potential (Han et al., 2025; Khairuddin et al., 2025; Loiodice et al., 2025; Zhou et al., 2025). These hypervirulent strains harbor multiple virulence genes, including rmpA, rmpA2, iucA, and iroB, which are intricately connected to their ability to adhere, mechanisms of iron acquisition, and capsule formation (Behera et al., 2025; Han et al., 2025; Loiodice et al., 2025; Wu et al., 2025). Specifically, the sequence type (ST) 23 of hvKP characteristically harbors virulence factors such as iucABCD, iroEN, and rmpA, facilitating the development of liver abscesses via advanced mechanisms of immune evasion and spread of the bacteria (Dhamelia et al., 2025; Khairuddin et al., 2025; Loiodice et al., 2025; Xin et al., 2025). It is important to note that hvKP strains have disseminated from Asia to Europe and the Americas, progressively emerging as a worldwide public health concern (Ko et al., 2002; Pope et al., 2011; Ejikeme et al., 2021). KP liver abscess (KPLA) represents a severe infection with a mortality rate of 3%–17% among hospitalized patients. If it progresses to invasive Klebsiella pneumoniae liver abscess syndrome (IKPLAS), the mortality rate surges to 20%–35%. IKPLAS is characterized by concurrent distant metastatic infections in the eyes, lungs, and central nervous system (CNS). This is accompanied by significantly increased intensive care unit admission rates and permanent visual or neurological disabilities, imposing a substantial burden. Therefore, IKPLAS has become an acute infection requiring urgent intervention (Zhang et al., 2022; Chang et al., 2023).

Early identification of high-risk patients is crucial for preventing IKPLAS. However, existing literature presents inconsistent and even contradictory conclusions regarding predictive risk factors for IKPLAS. This represents a bottleneck in clinical practice. A retrospective study in South Korea (Shin et al., 2013) has proposed that liver abscess size ≤ 5.8 cm and altered state of consciousness at admission are independent predictors of metastatic infection (Shin et al., 2013). A Chinese study focusing on imaging characteristics (Chang et al., 2015) has suggested that unilocular abscess and smaller abscess diameters are significantly associated with concurrent septic pulmonary embolism (Chang et al., 2015). Additionally, indicators, including diabetes, male gender, decreased platelet count, and sequential organ failure assessment (SOFA) score ≥ 4, have been individually validated by different cohorts. However, effect sizes vary significantly and are even contradictory (Li et al., 2020; Li et al., 2023; Gu et al., 2025; Sun et al., 2025). The fragmented evidence makes it difficult to establish unified, operable high-risk screening thresholds and early warning models for IKPLAS in clinical practice (Yoon et al., 2014; Wang et al., 2022).

To date, there have been no systematic reviews and meta-analyses to quantitatively integrate risk factors for IKPLAS. This study synthesized fragmented evidence to determine risk factors for IKPLAS. It aimed to provide scientific, precise, evidence-based medical support to reduce the burden of this fatal and disabling infection.

This study was conducted in conformity with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021). The study protocol has been registered in the international prospective systematic review registry PROSPERO (CRD420251143764).

This meta-analysis identified major risk factors for IKPLAS. The results revealed significant associations of the following factors with increased risk of IKPLAS: history of smoking or alcohol consumption, diabetes, liver disease, elevated SOFA score, septic shock, hepatic abscess of the left lobe, phlebitis, high PCT, high FBG, high PT, high total bilirubin, hypervirulent phenotype, and K1 serotype. No significant associations were found for other factors.

Diabetes and elevated FBG are recognized risk factors for IKPLAS, consistent with the study by Shin et al. in South Korea. In their study, diabetes patients accounted for 37.1% of all subjects, and the mean FBG at admission for all IKPLAS patients was 9.3 mmol/L, significantly higher than that in the non-metastasis group (6.1 mmol/L) (Wang et al., 2022). Diabetes, particularly with poor glycemic control, acts as a pivotal trigger that shifts Klebsiella pneumoniae from colonization to distant spread, resulting in invasive liver abscess. Chronic hyperglycemia remodels host-pathogen interactions through multiple mechanisms. First, advanced glycation end-products (AGEs) bind to their receptors (RAGE), activating NF-κB/NLRP3 inflammasomes and triggering cytokine storms. This causes damage to the hepatic sinusoidal endothelial barrier, opening a gateway for bacteria to enter the bloodstream. Meanwhile, the hyperosmolar microenvironment suppresses the phagocytic activity of Kupffer cells while diminishing the reactive oxygen species (ROS) burst in neutrophils and the formation of extracellular trap (NET), thus enabling bacteria to evade immune clearance. Furthermore, excessive glucose directly promotes the synthesis of capsular polysaccharide in K1-type hypervirulent strains via the phosphotransferase system (PTS), thus enhancing mucosal barriers and anti-phagocytic capabilities. Finally, the downregulation of insulin signaling reduces adiponectin, suppresses the activity of AMP-activated protein kinase (AMPK) in the liver, and impairs fatty acid oxidation, creating a lipid- and iron-rich environment for bacteria (Lin et al., 2013; Mohamed et al., 2016). However, although glycated hemoglobin reflects mean blood glucose levels over the preceding 8–12 weeks more stably and exhibits a linear association with the risk of invasiveness, its test is not routinely performed upon admission. In contrast, FBG is quickly accessible, inexpensive, and clinically feasible. Therefore, it is the preferred indicator for assessing uncontrolled blood glucose and predicting invasiveness in emergencies.

Smoking, alcohol consumption, and liver disease significantly increase the risk of IKPLAS. This is consistent with the predictive model in a study by Soochow University, which also identified liver disease as a key covariate for invasive infection in diabetic patients with KPLA (Feng et al., 2024). Nicotine in cigarettes inhibits the NF-κB pathway in alveolar macrophages, reducing IL-1β and TNF-α secretion and diminishing the initial clearance of Klebsiella pneumoniae (Chen et al., 2025). Harmful substances in smoke damage ciliated epithelial cells in the respiratory tract, impairing their ability to clear pathogens and leading to chronic bronchitis. This would compromise mucosal integrity, facilitating invasion of tissue by Klebsiella pneumoniae colonizing the respiratory tract (Zhang et al., 2021). Alcohol directly destroys intestinal tight junctions and elevates portal vein endotoxin levels. Consequently, hepatic Kupffer cells are activated to produce IL-10, creating an immunotolerant microenvironment (Chen et al., 2025). Acetaldehyde, a metabolite of ethanol, inhibits NADPH oxidase activity in neutrophils, reduces ROS-mediated bactericidal effects, and promotes synergistic immune evasion with the hypermucoviscous phenotype of KP (Mohamed et al., 2016). Therefore, smoking, alcohol consumption, and liver disease collectively amplify the invasive potential of KP through three pathways: mucosal barrier damage, immune deviation, and microenvironment hypoxia. Clinically, these three factors should be incorporated into the early warning scoring criteria for IKPLAS.

Elevated SOFA scores and septic shock are independent predictors for IKPLAS, which aligns with the conclusion of Sun et al.’s cohort study on multicenter community-acquired KP invasive infection conducted in Guangdong, China (Sun et al., 2025). Their study has demonstrated that each 1-point increase in SOFA score at admission is associated with an increase in mortality risk by 16%, with 61.9% of patients with KP invasive infection developing septic shock. The two studies collectively suggest that the key determinant of invasion and dissemination is multi-organ dysfunction reflected by SOFA, rather than the size of the hepatic lesion alone. Elevated SOFA scores are often accompanied by cytokine storms, that is, marked increases in IL-6 and TNF-α levels. This could upregulate ICAM-1 and EGFR in hepatocytes and bile duct epithelial cells, promoting adhesion and transcytosis of hypermucoviscous KP (Serban et al., 2021). Concurrently, the imbalance of coagulation and fibrinolysis (decreased platelet count and increased international normalized ratio) increases microthrombi in the hepatic sinusoid, leading to ischemia and necrosis of the abscess wall. This creates a “pressure valve” effect, allowing bacteria to enter the systemic circulation via the short hepatic vein-inferior vena cava pathway and induce shock. During shock, increased systemic vascular permeability further enhances the probability of distant migration (eyes, brain, lungs), forming a positive feedback loop. Although septic shock was identified as a significant risk factor, this finding was based on only 2 studies. The small sample size may lead to instability of effect estimates. Hence, the results should be interpreted with caution, and this risk factor needs to be further studied utilizing expanded literature. The cohort study in Hebei, China, though with a small sample size, has also demonstrated a strong association of high SOFA scores with gas formation in abscesses and decreased neutrophil ratios. This suggests that rapid gas production from necrotic liver parenchyma accelerates organ failure (Zhang et al., 2024). Therefore, in any patient with KPLA, a SOFA score ≥ 4 or early signs of shock should be regarded as a warning signal of progression to invasive KPLA. Particularly in patients with underlying conditions, proactive screening for hepatic lesions and intensified anti-infective therapy are essential to block this vicious cycle.

Phlebitis is the strongest and stablest independent predictor of IKPLAS, which strongly aligns with multicenter findings from studies of Gu et al., Li et al., and Wang et al (Wang et al., 2022; Li et al., 2023; Gu et al., 2025). Its mechanism transcends simple thrombotic occlusion: the hypermucoviscous KP activates TLR2/4 via magA/rmpA, elevating the expression of ICAM-1 and TF on vascular endothelium to cause in situ immunothrombosis (Lee et al., 2010). Bacterial capsular polysaccharides further shield the bacteria from neutrophil phagocytosis, enabling continuous release of bacterial emboli within the thrombus. These emboli disseminate along the portal-systemic-pulmonary axis or the hepatic vein-inferior vena cava-right heart-pulmonary circulation pathway, significantly increasing the rates of metastasis to the lungs, eyes, and CNS. The left hepatic lobe abscess is also frequently listed as a risk factor in multiple studies due to the anatomical characteristics of the lobe. The left hepatic vein branches are short with large diameters and share a common trunk with the middle and left hepatic veins directly connecting the inferior vena cava. Elevated abscess cavity pressure makes it easier for bacterial emboli to enter the systemic circulation. Besides, the left hepatic lobe and the left branch of the portal vein run in a straight line. This allows bacterial emboli in the portal vein to rapidly reach the splenic vein and superior mesenteric vein, inducing distant infections in the pancreas, spleen, and kidneys (Lee et al., 2005). Thus, phlebitis reflects the dissemination pathway formed by the interaction of the bacteria-vessel-immune triad, while left hepatic lobe abscess provides a geographical shortcut. The synergistic effect of both significantly increases the probability of IKPLAS.

Elevated levels of PCT, PT, and total bilirubin all increase the invasive potential of KPLA, which aligns with Sun et al.’s findings based on 291 cases of community-acquired infections (Sun et al., 2025). However, they only regard PCT as a marker of inflammation severity, without explaining why high PCT predicts distant dissemination. This can be explained by integrating the results of this meta-analysis with the mechanism study. When PCT > 10 ng/mL, the lactate concentration in the liver abscess microenvironment increases simultaneously, inhibiting cAMP via the PTS-CRP axis. This triggers high expression of capsular polysaccharide (CPS), enhancing bacterial resistance to phagocytosis and facilitating penetration of the hepatic sinusoids (Feng et al., 2023). Meanwhile, prolonged PT reflects activation of the KP endotoxin-TLR4 pathway, inducing extrinsic coagulation. Fibrin deposition forms bacterial microthrombi (Zhu et al., 2025), which shield bacteria from phagocytosis and embolize target organs such as the eyes and brain via the bloodstream. More importantly, total bilirubin > 20 μmol/L is often accompanied by compression of the bile duct or abscess rupture into the biliary tract. Decreased glycohyocholic acid concentration in the bile weakens its negative regulation of CPS synthesis, while conjugated bilirubin itself induces IL-1β release, amplifying the inflammatory cascade and providing an inflammatory soil for distant colonization of bacteria (Li et al., 2023). Thus, these three indicators are not merely passive outcomes of severe conditions but actively participate in the invasion process through the lactate-capsule axis, microthrombosis-dissemination axis, and bile-immune axis. Clinically, simultaneous elevation of PCT, PT, and total bilirubin should be regarded as a high-risk invasion window. In addition to adequate drainage and sensitive antimicrobial therapy, early management of potential migration sites and intervention in coagulation and cholestasis are expected to reduce distant infection rates.

The hypervirulent (HV) phenotype and K1 serotype are independent risk factors for IKPLAS, which aligns with Sun et al.’s findings from 291 cases of community-acquired infections in Guangdong, China (Sun et al., 2025). Lee et al. (2010) have further reported that HV-positive strains constitute 94.7% of IKPLAS strains, with 91.4% carrying rmpA/rmpA2. This suggests that rmpA activates the extracellular polysaccharide synthesis operon, enabling colonies to form > 10 mm mucous filaments to evade Kupffer cell phagocytosis in hepatic sinusoids and promote hematogenous dissemination. The K1 capsules inhibit complement C3 deposition and neutrophil chemotaxis because they contain magA-mediated 2,3-dihydro-2-deoxy-N-acetylmannuronic acid polymers, whose concentrations were positively associated with IL-6 level and SOFA score. Zhang et al. (2022) have identified significantly higher magA abundance in IKPLAS isolates with K1 serotype compared to non-invasive strains, accompanied by upregulation of iron transporter genes iucA and fepC. This indicates that K1 capsules and iron uptake synergistically enhance the tolerance to oxidative stress, which accelerates extrahepatic migration. Driven by advances in whole-genome and comparative genomics techniques, high-throughput sequencing has progressively revealed the virulence gene clusters, synthesis pathways of CPS, and essential regulatory elements associated with the HV phenotype. These findings establish a molecular basis for targeted intervention strategies (Liu et al., 2025). The HV phenotype and the K1 CPS serve as pivotal virulence markers characteristic of IKPLAS. The rmpA/rmpA2 genes activate the CPS cluster to synthesize a high-molecular-weight capsule (> 40 kDa) that impedes neutrophil phagocytosis. The magA gene facilitates intracellular transit, assisting in the bacterial translocation from the intestine to the liver via the portal vein. Strains of the K1 serotype utilize a dual iron transport system comprising aerobactin and iroBCDN, the expression of which is significantly increased in high-glucose conditions. This system captures host transferrin, leading to an NF-κB/IL-6 induced inflammatory cascade and the formation of multiple liquefied abscesses (Lee et al., 2009; Tu et al., 2024). The coexistence of diabetes, K1 serotype, and high levels of mucus elevates the risk of IKPLAS, indicating the co-selection of high virulence and host metabolism. This provides evidence to guide the precise control of infection and the development of targeted anti-virulence therapies.

It is noteworthy that abdominal symptoms (e.g., abdominal pain) are considered protective factors. When such symptoms are present, greater attention is paid to the risk of secondary infection during treatment, thereby preventing IKPLAS to some extent. Additionally, an elevated albumin level is also considered a protective factor. However, since this meta-analysis incorporated a limited number of studies, it is necessary to further validate the reliability of this conclusion.

This study may provide references for future treatment of patients with KPLA, assisting clinicians in preventing invasive infections. Clinicians should be watchful and implement preventive measures for patients with a history of smoking or alcohol consumption, especially those with diabetes, liver disease, elevated SOFA scores, septic shock, phlebitis, or left hepatic lobe abscesses. Precautions are also warranted for patients exhibiting abnormal laboratory indicators such as elevated PCT, FBG, PT, or total bilirubin, or those with isolates of the HV phenotype or K1serotype.

Among the variables involved in this study, marked heterogeneity is observed in 8 variables, including hypertension, SOFA, liver abscess size, simple leaf, septa, abnormal perfusion, glycated hemoglobin, and PLT. This diminishes the accuracy of pooled effect estimates. Additionally, 4 studies incorporated in this meta-analysis are rated medium-quality, which may also introduce potential bias and compromise the credibility of the results.

This study represents the first comprehensive meta-analysis of risk factors for IKPLAS through an extensive literature review. However, it has some limitations. First, a small number of studies are included. Some indicators are reported in only a few studies. This may affect the external validity and accuracy of the results. Second, publication bias is present. Gender-related publication bias is observed, reducing the reliability of results and necessitating further validation. Additionally, geographical limitations exist. The included studies are mainly from China and South Korea, suggesting that the identified risk factors may be more applicable to Asian populations. These limitations partially undermine the credibility of the conclusion of this study. Future studies should incorporate samples from more countries and regions to enhance the general applicability of identified risk factors. For instance, international multi-center studies covering Europe, North America, and South America should be carried out. Rigorously designed, large-scale prospective cohort studies are necessary to confirm the causal relationships between risk factors and invasive infections and to further validate the general applicability of the findings.

This study is the first comprehensive meta-analysis to identify key risk factors contributing to the progression of KPLA to IKPLAS: smoking, alcohol consumption, diabetes, underlying liver disease, elevated SOFA scores, septic shock, phlebitis, left hepatic lobe abscess, as well as elevated PCT, FBG, PT, or total bilirubin levels, and infection with strains of K1 serotype. Future large-scale prospective cohort studies should be conducted in expanded geographical regions to establish an evidence-based platform for precise early warning and intervention.