医学
肝移植
CXCL14型
内科学
生物标志物
对乙酰氨基酚
胃肠病学
移植
趋化因子
药理学
趋化因子受体
炎症
生物化学
化学
作者
David S. Umbaugh,Nga Nguyen,Steven C. Curry,Jody A. Rule,William M. Lee,Anup Ramachandran,Hartmut Jaeschke
标识
DOI:10.1097/hep.0000000000000665
摘要
Background and Aims: Patients with acetaminophen-induced acute liver failure are more likely to die while on the liver transplant waiting list than those with other causes of acute liver failure. Therefore, there is an urgent need for prognostic biomarkers that can predict the need for liver transplantation early after an acetaminophen overdose. Approach and Results: We evaluated the prognostic potential of plasma chemokine C-X-C motif ligand 14 (CXCL14) concentrations in patients with acetaminophen (APAP) overdose (n=50) and found that CXCL14 is significantly higher in nonsurviving patients compared to survivors with acute liver failure ( p < 0.001). Logistic regression and AUROC analyses revealed that CXCL14 outperformed the MELD score, better discriminating between nonsurvivors and survivors. We validated these data in a separate cohort of samples obtained from the Acute Liver Failure Study Group (n = 80), where MELD and CXCL14 had similar AUC (0.778), but CXCL14 demonstrated higher specificity (81.2 vs. 52.6) and positive predictive value (82.4 vs. 65.4) for death or need for liver transplantation. Next, combining the patient cohorts and using a machine learning training/testing scheme to mimic the clinical scenario, we found that CXCL14 outperformed MELD based on AUC (0.821 vs. 0.787); however, combining MELD and CXCL14 yielded the best AUC (0.860). Conclusions: We find in 2 independent cohorts of acetaminophen overdose patients that circulating CXCL14 concentration is a novel early prognostic biomarker for poor outcomes, which may aid in guiding decisions regarding patient management. Moreover, our findings reveal that CXCL14 performs best when measured soon after patient presentation to the clinic, highlighting its importance for early warning of poor prognosis.
科研通智能强力驱动
Strongly Powered by AbleSci AI