Risk evaluation of carbapenem-induced liver injury based on machine learning analysis

Drug-induced liver injury (DILI) is a severe adverse drug reaction; idiosyncratic DILI accounts for 11% of all DILI cases [1]. Many studies on DILI have reported antibiotics as the most common causative agents [2]. Carbapenems are broad-spectrum antibiotics used to treat sepsis and septic shock caused by antimicrobial-resistant bacteria, such as AmpC β-Lactamase-producing Enterobacter spp [3] and expended β-Lactamase-producing bacteria [4]. Data mining analysis has shown that meropenem (MEPM), a carbapenem, is strongly associated with DILI [5,6], suggesting that liver function should be frequently monitored during its administration. Doripenem (DRPM) is a relatively new carbapenem whose antimicrobial spectrum closely resembles that of MEPM [7]. DRPM is active against antimicrobial-resistant bacteria associated with hospital-acquired pneumonia [8] and febrile neutropenia [9]. Recently, a network meta-analysis reported that the pattern of adverse drug reactions might differ between MEPM and DRPM [10], suggesting that the occurrence rate of DILI in MEPM and DRPM may differ. However, there is limited comparative information regarding DILI from these agents because DRPM is not readily used in clinical settings when compared with MEPM.

Patients with severe infections are treated with carbapenems, suggesting that appropriate management of carbapenem-induced liver injury is cardinal in preventing patient mortality. Because DILI tends to be frequently missed and overlooked in clinical settings, its incidence rate is estimated to be low [11]. Thus, to the best of our knowledge, there is currently no evidence on carbapenem-induced liver injury risk factors. In addition, because patients have multiple risk factors, including age, liver function, and underlying diseases in clinical settings, the combination of these factors requires a comprehensive assessment to understand the relationships with DILI.

Decision tree (DT) analysis, a type of machine learning, consists of a flowchart-like model in which users can easily assess the risk of adverse drug reactions by combining multiple risk factors [[12], [13], [14]]. Therefore, DT analysis may be a valuable assessment tool for DILI in carbapenems administered by medical staff; however, this remains to be explored.

In this study, we aimed to compare the DILI rate between MEPM and DRPM and construct a simple flowchart that can be used to predict liver injury before initiating carbapenem therapy.

留言 (0)

沒有登入
gif