ObjectiveTumour pathology contains rich information, including tissue structure and cell morphology, that reflects disease progression and patient survival. However, phenotypic information is subtle and complex, making the discovery of prognostic indicators from pathological images challenging.DesignAn interpretable, weakly supervised deep learning framework incorporating prior knowledge was proposed to analyse hepatocellular carcinoma (HCC) and explore new prognostic phenotypes on pathological whole-slide images (WSIs) from the Zhongshan cohort of 1125 HCC patients (2451 WSIs) and TCGA cohort of 320 HCC patients (320 WSIs). A ‘tumour risk score (TRS)’ was established to evaluate patient outcomes, and then risk activation mapping (RAM) was applied to visualise the pathological phenotypes of TRS. The multi-omics data of The Cancer Genome Atlas(TCGA) HCC were used to assess the potential pathogenesis underlying TRS.ResultsSurvival analysis revealed that TRS was an independent prognosticator in both the Zhongshan cohort (p<0.0001) and TCGA cohort (p=0.0003). The predictive ability of TRS was superior to and independent of clinical staging systems, and TRS could evenly stratify patients into up to five groups with significantly different prognoses. Notably, sinusoidal capillarisation, prominent nucleoli and karyotheca, the nucleus/cytoplasm ratio and infiltrating inflammatory cells were identified as the main underlying features of TRS. The multi-omics data of TCGA HCC hint at the relevance of TRS to tumour immune infiltration and genetic alterations such as the FAT3 and RYR2 mutations.ConclusionOur deep learning framework is an effective and labour-saving method for decoding pathological images, providing a valuable means for HCC risk stratification and precise patient treatment.

Cardiac paraganglioma is a rare entity. We review the clinical data from 158 patients reported in 132 isolated papers, and discuss clinical presentations, imaging findings, pathology, location, therapy, and outcomes.

The comprehensive assessment indicates that autonomic dysfunction occurs in acute phase of ischemic stroke and may persist up to six-months after stroke. Parasympathetic dysfunction rather than sympathetic dysfunction is predominant after ischemic stroke.

Until recently, 10% of hepatocellular adenomas (HCAs) remained unclassified (UHCA). Among the UHCAs, the sonic hedgehog HCA (shHCA) was defined by focal deletions that fuse the promoter of Inhibin beta E chain with GLI1. Prostaglandin D2 synthase was proposed as immunomarker. In parallel, our previous work using proteomic analysis showed that most UHCAs constitute a homogeneous subtype associated with overexpression of argininosuccinate synthase (ASS1). To clarify the use of ASS1 in the HCA classification and avoid misinterpretations of the immunohistochemical staining, the aims of this work were to study (1) the link between shHCA and ASS1 overexpression and (2) the clinical relevance of ASS1 overexpression for diagnosis. Molecular, proteomic, and immunohistochemical analyses were performed in UHCA cases of the Bordeaux series. The clinico-pathological features, including ASS1 immunohistochemical labeling, were analyzed on a large international series of 67 cases. ASS1 overexpression and the shHCA subgroup were superimposed in 15 cases studied by molecular analysis, establishing ASS1 overexpression as a hallmark of shHCA. Moreover, the ASS1 immunomarker was better than prostaglandin D2 synthase and only found positive in 7 of 22 shHCAs. Of the 67 UHCA cases, 58 (85.3%) overexpressed ASS1, four cases were ASS1 negative, and in five cases ASS1 was noncontributory. Proteomic analysis performed in the case of doubtful interpretation of ASS1 overexpression, especially on biopsies, can be a support to interpret such cases. ASS1 overexpression is a specific hallmark of shHCA known to be at high risk of bleeding. Therefore, ASS1 is an additional tool for HCA classification and clinical diagnosis.