Sorbonne Université, Assistance Publique-Hôpitaux de Paris, Hôpital Pitié Salpêtrière, Institute of Cardiometabolism and Nutrition (ICAN), Paris, France.
Antwerp University Hospital, Antwerp, Belgium.
Hepatology. 2024 Jul 1;80(1):173-185. doi: 10.1097/HEP.0000000000000723. Epub 2023 Dec 19.
Artificial intelligence-powered digital pathology offers the potential to quantify histological findings in a reproducible way. This analysis compares the evaluation of histological features of NASH between pathologists and a machine-learning (ML) pathology model.
This post hoc analysis included data from a subset of patients (n=251) with biopsy-confirmed NASH and fibrosis stage F1-F3 from a 72-week randomized placebo-controlled trial of once-daily subcutaneous semaglutide 0.1, 0.2, or 0.4 mg (NCT02970942). Biopsies at baseline and week 72 were read by 2 pathologists. Digitized biopsy slides were evaluated by PathAI's NASH ML models to quantify changes in fibrosis, steatosis, inflammation, and hepatocyte ballooning using categorical assessments and continuous scores. Pathologist and ML-derived categorical assessments detected a significantly greater percentage of patients achieving the primary endpoint of NASH resolution without worsening of fibrosis with semaglutide 0.4 mg versus placebo (pathologist 58.5% vs. 22.0%, p < 0.0001; ML 36.9% vs. 11.9%; p =0.0015). Both methods detected a higher but nonsignificant percentage of patients on semaglutide 0.4 mg versus placebo achieving the secondary endpoint of liver fibrosis improvement without NASH worsening. ML continuous scores detected significant treatment-induced responses in histological features, including a quantitative reduction in fibrosis with semaglutide 0.4 mg versus placebo ( p =0.0099) that could not be detected using pathologist or ML categorical assessment.
ML categorical assessments reproduced pathologists' results of histological improvement with semaglutide for steatosis and disease activity. ML-based continuous scores demonstrated an antifibrotic effect not measured by conventional histopathology.
人工智能驱动的数字病理学具有以可重复的方式定量评估组织学发现的潜力。本分析比较了病理学家和机器学习（ML）病理学模型对 NASH 组织学特征的评估。
本回顾性分析纳入了来自一项为期 72 周、随机、安慰剂对照的每日一次皮下司美格鲁肽 0.1、0.2 或 0.4mg 治疗非酒精性脂肪性肝炎（NASH）和纤维化分期 F1-F3的亚组患者（n=251）的数据（NCT02970942）。基线和第 72 周的活检由 2 名病理学家进行解读。PathAI 的 NASH ML 模型对数字化活检切片进行评估，使用分类评估和连续评分来量化纤维化、脂肪变性、炎症和肝细胞气球样变的变化。病理学家和 ML 衍生的分类评估检测到接受司美格鲁肽 0.4mg 治疗的患者与安慰剂相比，达到 NASH 缓解且纤维化无恶化的主要终点的比例显著更高（病理学家 58.5% vs. 22.0%，p<0.0001；ML 36.9% vs. 11.9%；p=0.0015）。两种方法均检测到接受司美格鲁肽 0.4mg 治疗的患者与安慰剂相比，达到次要终点肝纤维化改善且 NASH 无恶化的比例更高，但无统计学意义。ML 连续评分检测到组织学特征的显著治疗反应，包括司美格鲁肽 0.4mg 与安慰剂相比纤维化定量减少（p=0.0099），这一结果无法通过病理学家或 ML 分类评估检测到。
ML 分类评估再现了病理学家使用司美格鲁肽治疗脂肪变性和疾病活动的组织学改善结果。基于 ML 的连续评分显示了一种抗纤维化作用，这是常规组织病理学无法测量的。
