Risk stratification after paracetamol overdose using mechanistic biomarkers: results from two prospective cohort studies



Dear, James W, Clarke, Joanna I, Francis, Ben ORCID: 0000-0002-2130-5976, Allen, Lowri, Wraight, Jonathan, Shen, Jasmine, Dargan, Paul I, Wood, David, Cooper, Jamie, Thomas, Simon HL
et al (show 4 more authors) (2018) Risk stratification after paracetamol overdose using mechanistic biomarkers: results from two prospective cohort studies. The Lancet Gastroenterology and Hepatology, 3 (2). 104 - 113.

[img] Text
C:\Users\munirp\Dropbox (Personal)\own articles\2017\lancet gastro hepatology.pdf - Published Version

Download (186kB)

Abstract

Background Paracetamol overdose is common but patient stratification is suboptimal. We investigated the usefulness of new biomarkers that have either enhanced liver specificity (microRNA-122 [miR-122]) or provide mechanistic insights (keratin-18 [K18], high mobility group box-1 [HMGB1], and glutamate dehydrogenase [GLDH]). The use of these biomarkers could help stratify patients for their risk of liver injury at hospital presentation. Methods Using data from two prospective cohort studies, we assessed the potential for biomarkers to stratify patients who overdose with paracetamol. We completed two independent prospective studies: a derivation study (MAPP) in eight UK hospitals and a validation study (BIOPAR) in ten UK hospitals. Patients in both cohorts were adults (≥18 years in England, ≥16 years in Scotland), were diagnosed with paracetamol overdose, and gave written informed consent. Patients who needed intravenous acetylcysteine treatment for paracetamol overdose had circulating biomarkers measured at hospital presentation. The primary endpoint was acute liver injury indicating need for continued acetylcysteine treatment beyond the standard course (alanine aminotransferase [ALT] activity >100 U/L). Receiver operating characteristic (ROC) curves, category-free net reclassification index (cfNRI), and integrated discrimination index (IDI) were applied to assess endpoint prediction. Findings Between June 2, 2010, and May 29, 2014, 1187 patients who required acetylcysteine treatment for paracetamol overdose were recruited (985 in the MAPP cohort; 202 in the BIOPAR cohort). In the derivation and validation cohorts, acute liver injury was predicted at hospital presentation by miR-122 (derivation cohort ROC–area under the curve [AUC] 0·97 [95% CI 0·95–0·98]), HMGB1 (0·95 [0·93–0·98]), and full-length K18 (0·95 [0·92–0·97]). Results were similar in the validation cohort (miR-122 AUC 0·97 [95% CI 0·95–0·99], HMGB1 0·98 [0·96–0·99], and full-length K18 0·93 [0·86–0·99]). A combined model of miR-122, HMGB1, and K18 predicted acute liver injury better than ALT alone (cfNRI 1·95 [95% CI 1·87–2·03], p<0·0001 in the MAPP cohort; 1·54 [1·08–2·00], p<0·0001 in the BIOPAR cohort). Interpretation Personalised treatment pathways could be developed by use of miR-122, HMGB1, and full-length K18 at hospital presentation for patient stratification. This prospective study supports their use for hepatic safety assessment of new medicines.

Item Type: Article
Depositing User: Symplectic Admin
Date Deposited: 20 Nov 2017 15:32
Last Modified: 28 May 2020 12:10
DOI: 10.1016/S2468-1253(17)30266-2
Related URLs:
URI: http://livrepository.liverpool.ac.uk/id/eprint/3012559