Anatomical, histomorphological and molecular classification of cholangiocarcinoma

Visual inspection of histopathology slides is one of the main methods used by pathologists to assess the stage, type and subtype of lung tumors. Adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC) are the most prevalent subtypes of lung cancer, and their distinction requires visual inspection by an experienced pathologist. In this study, we trained a deep convolutional neural network (inception v3) on whole-slide images obtained from The Cancer Genome Atlas to accurately and automatically classify them into LUAD, LUSC or normal lung tissue. The performance of our method is comparable to that of pathologists, with an average area under the curve (AUC) of 0.97. Our model was validated on independent datasets of frozen tissues, formalin-fixed paraffin-embedded tissues and biopsies. Furthermore, we trained the network to predict the ten most commonly mutated genes in LUAD. We found that six of them-STK11, EGFR, FAT1, SETBP1, KRAS and TP53-can be predicted from pathology images, with AUCs from 0.733 to 0.856 as measured on a held-out population. These findings suggest that deep-learning models can assist pathologists in the detection of cancer subtype or gene mutations. Our approach can be applied to any cancer type, and the code is available at https://github.com/ncoudray/DeepPATH .

Summary Cholangiocarcinoma (CCA) is an aggressive malignancy of the bile ducts, with poor prognosis and limited treatment options. Here, we describe the integrated analysis of somatic mutations, RNA expression, copy number, and DNA methylation by The Cancer Genome Atlas of a set of predominantly intrahepatic CCA cases and propose a molecular classification scheme. We identified an IDH mutant-enriched subtype with distinct molecular features including low expression of chromatin modifiers, elevated expression of mitochondrial genes, and increased mitochondrial DNA copy number. Leveraging the multi-platform data, we observed that ARID1A exhibited DNA hypermethylation and decreased expression in the IDH mutant subtype. More broadly, we found that IDH mutations are associated with an expanded histological spectrum of liver tumors with molecular features that stratify with CCA. Our studies reveal insights into the molecular pathogenesis and heterogeneity of cholangiocarcinoma and provide classification information of potential therapeutic significance.

Cholangiocarcinoma is an intractable cancer, with limited therapeutic options, in which the molecular mechanisms underlying tumor development remain poorly understood. Identification of a novel driver oncogene and applying it to targeted therapies for molecularly defined cancers might lead to improvements in the outcome of patients. We performed massively parallel whole transcriptome sequencing in eight specimens from cholangiocarcinoma patients without KRAS/BRAF/ROS1 alterations and identified two fusion kinase genes, FGFR2-AHCYL1 and FGFR2-BICC1. In reverse-transcriptase polymerase chain reaction (RT-PCR) screening, the FGFR2 fusion was detected in nine patients with cholangiocarcinoma (9/102), exclusively in the intrahepatic subtype (9/66, 13.6%), rarely in colorectal (1/149) and hepatocellular carcinoma (1/96), and none in gastric cancer (0/212). The rearrangements were mutually exclusive with KRAS/BRAF mutations. Expression of the fusion kinases in NIH3T3 cells activated MAPK and conferred anchorage-independent growth and in vivo tumorigenesis of subcutaneous transplanted cells in immune-compromised mice. This transforming ability was attributable to its kinase activity. Treatment with the fibroblast growth factor receptor (FGFR) kinase inhibitors BGJ398 and PD173074 effectively suppressed transformation.