DataSheet1_The LAC Score Indicates Significant Fibrosis in Patients With Chronic Drug-Induced Liver Injury: A Large Biopsy-Based Study.docx

Currently, there are no satisfactory noninvasive methods for the diagnosis of fibrosis in patients with chronic drug-induced liver injury (DILI). Our goal was to develop an algorithm to improve the diagnostic accuracy of significant fibrosis in this population. In the present study, we retrospectively investigated the biochemical and pathological characteristics of consecutive patients with biopsy-proven chronic DILI, who presented at our hospital from January 2013 to December 2017. A noninvasive algorithm was developed by using multivariate logistic regression, receiver operating characteristic (ROC) curves, and decision curve analysis (DCA) to diagnose significant fibrosis in the training cohort, and the algorithm was subsequently validated in the validation cohort. Totally, 1,130 patients were enrolled and randomly assigned into a training cohort (n = 848) and a validation cohort (n = 282). Based on the multivariate analysis, LSM, CHE, and APRI were independently associated with significant fibrosis. A novel algorithm, LAC, was identified with the AUROC of 0.81, which was significantly higher than LSM (AUROC 0.78), CHE (AUROC 0.73), and APRI (AUROC 0.68), alone. The best cutoff value of LAC in the training cohort was 5.4. When the LAC score was used to diagnose advanced fibrosis and cirrhosis stages, the optimal cutoff values were 6.2 and 6.7, respectively, and the AUROC values were 0.84 and 0.90 in the training cohort and 0.81 and 0.83 in the validation cohort. This study proved that the LAC score can contribute to the accurate assessment of high-risk disease progression and the establishment of optimal treatment strategies for patients with chronic DILI.

目前，针对慢性药物性肝损伤（drug-induced liver injury, DILI）患者的肝纤维化诊断，尚无令人满意的无创检测手段。本研究旨在构建一种算法，以提升该人群中显著肝纤维化的诊断准确度。本研究回顾性分析了2013年1月至2017年12月于我院就诊的连续纳入的、经活检证实的慢性DILI患者的生化与病理学特征。本研究通过多因素logistic回归、受试者工作特征（receiver operating characteristic, ROC）曲线及决策曲线分析（decision curve analysis, DCA）构建了用于诊断显著肝纤维化的无创算法，并在验证队列中对该算法进行了验证。本研究共纳入1130例患者，按随机分配原则分为训练队列（n=848）与验证队列（n=282）。经多因素分析，肝脏硬度测量值（liver stiffness measurement, LSM）、胆碱酯酶（cholinesterase, CHE）及天冬氨酸转氨酶与血小板比值指数（aspartate aminotransferase-to-platelet ratio index, APRI）均与显著肝纤维化独立相关。本研究构建了一种名为LAC的新型算法，其受试者工作特征曲线下面积（area under the receiver operating characteristic curve, AUROC）为0.81，显著高于单一使用LSM（AUROC=0.78）、CHE（AUROC=0.73）及APRI（AUROC=0.68）的诊断效能。在训练队列中，LAC评分的最佳截断值为5.4。当使用LAC评分诊断进展期肝纤维化及肝硬化分期时，其最佳截断值分别为6.2与6.7；在训练队列中对应的AUROC分别为0.84与0.90，在验证队列中则分别为0.81与0.83。本研究证实，LAC评分可帮助准确评估慢性DILI患者的高危疾病进展风险，并为其制定最优治疗策略提供依据。