Development and validation of a predictive model for invasive syndrome in patients with Klebsiella pneumoniae liver abscess.

Introduction: Pyogenic liver abscess (PLA) is a life-threatening liver bacterial infection causing suppurative lesions. In Asia, hypervirulent Klebsiella pneumoniae (hvKP) is the main PLA pathogen, linked to invasive syndromes. Severe Klebsiella pneumoniae liver abscess (KPLA) manifestations, called invasive KPLA syndrome (IKPLAS), have acute onset, rapid progression and non-specific symptoms, often leading to poor prognosis if untreated. This study aimed to find risk factors and create a validated nomogram for predicting invasive syndrome in KPLA patients.

Methods: We retrospectively analyzed clinical data from 529 KPLA patients treated at Chengde Medical University Affiliated Hospital between August 1, 2014, and November 30, 2024. By using the 7:3 stratified random sampling method, the patients were assigned to two cohorts: derivation (n = 370) and validation (n = 159). Univariate and multivariate logistic regression analyses were performed to identify risk factors for invasive KPLA syndrome (IKPLAS). A predictive nomogram was constructed and evaluated for discrimination and clinical utility.

Results: Of the 529 enrolled patients, 33 patients (6.2%) developed IKPLAS (IKPLAS group), while the remaining 496 patients were included in the non-invasive group (NIKPLAS group). Both groups showed significant differences (P < 0.05) in the incidence of viral hepatitis, biliary disease, type 2 diabetes mellitus (T2DM), vomiting, pulmonary infection, and septic shock; C-reactive protein level; abscess diameter; presence of a gas-containing abscess; and Sequential Organ Failure Assessment (SOFA) score. Multivariate analysis identified the following factors as independent predictors: viral hepatitis, T2DM, abscess diameter, presence of a gas-containing abscess, and SOFA score. The nomogram showed excellent calibration (Hosmer-Lemeshow χ² = 4.171, P = 0.841) with area under the receiver operating characteristic curve values of 0.961 (derivation cohort) and 0.899 (validation cohort). The clinical utility of the nomogram was confirmed by decision curve analysis.

Conclusion: Viral hepatitis, T2DM, abscess diameter, presence of a gas-containing abscess, and SOFA score are the predictive factors of IKPLAS. The developed nomogram provides reliable risk stratification for patients with KPLA and can be applied clinically to predict IKPLAS cases.