Transition between differentiation states in development occurs swift but the mechanisms leading to epigenetic and transcriptional reprogramming are poorly understood. The pediatric cancer neuroblastoma includes adrenergic (ADRN) and mesenchymal (MES) tumor cell types, which differ in phenotype, super-enhancers (SEs) and core regulatory circuitries. These cell types can spontaneously interconvert, but the mechanism remains largely unknown. Here, we unravel how a NOTCH3 intracellular domain reprogrammed the ADRN transcriptional landscape towards a MES state. A transcriptional feed-forward circuitry of NOTCH-family transcription factors amplifies the NOTCH signaling levels, explaining the swift transition between two semi-stable cellular states. This transition induces genome-wide remodeling of the H3K27ac landscape and a switch from ADRN SEs to MES SEs. Once established, the NOTCH feed-forward loop maintains the induced MES state. In vivo reprogramming of ADRN cells shows that MES and ADRN cells are equally oncogenic. Our results elucidate a swift transdifferentiation between two semi-stable epigenetic cellular states.

Among breast cancer patients who develop distant metastases, there is marked variability in the clinical course, including metastasis pattern. Here, we present a retrospective study of breast cancer patients who all developed distant metastases focusing on the association between breast cancer subtype and clinical course, including organ-specific metastasis. Tissue microarrays (TMAs) were assembled and stained for ER, PR, HER2, EGFR, CK5/6, CK14, E-Cadherin, TP53 and Ki67 for 263 breast cancer patients with metastatic disease. Tumours were classified into ER+/HER2−/Ki67high, ER+/HER2−/Ki67low, ER+/HER2+, ER−/HER2+ and ER−/HER2− groups. Relevant data related to metastasis pattern, metastasis timeline, systemic treatment and survival were retrieved. Associations between site-specific relapse and patient/tumour characteristics were assessed with multivariate models using logistic regression. Median time for development of distant metastasis was 30 months (range 0–15.3 years); 75.8 % of the distance metastases developed in the first 5 years after treatment of the primary tumour. Patients with ER−/HER2− tumours had a median overall survival of 27 months; those with HER2+ tumours of 52 months; those with ER+/HER2−/Ki67high of 76 months and those with ER+/HER2−/Ki67low of 79 months. Bone was the most common site for distant metastasis (70.6 %) followed by liver (54.5 %) and lung (31.4 %), respectively. Visceral metastasis was found in 76.8 % of the patients. Patients with ER−/HER2− tumours developed visceral metastases in 81 % and bone metastases in 55.2 %; those with HER2+ tumours developed visceral metastases in 77.4 % and bone metastases in 69.8 %; those with ER+/HER2−/Ki67high developed visceral metastases in 75.7 % and bone metastases in 87.8 % and those with ER+/HER2−/Ki67low developed visceral metastases in 76.9 % and bone metastases in 73.1 %. In metastatic breast cancer patients, tumour subtypes are associated with survival and pattern of distant metastases. These associations are of help in choices for surveillance and therapy in individual patients.

Histopathologic grading of prostate cancer using Gleason patterns (GPs) is subject to a large inter-observer variability, which may result in suboptimal treatment of patients. With the introduction of digitization and whole-slide images of prostate biopsies, computer-aided grading becomes feasible. Computer-aided grading has the potential to improve histopathological grading and treatment selection for prostate cancer. Automated detection of GPs and determination of the grade groups (GG) using a convolutional neural network. In total, 96 prostate biopsies from 38 patients are annotated on pixel-level. Automated detection of GP 3 and GP ≥ 4 in digitized prostate biopsies is performed by re-training the Inception-v3 convolutional neural network (CNN). The outcome of the CNN is subsequently converted into probability maps of GP ≥ 3 and GP ≥ 4, and the GG of the whole biopsy is obtained according to these probability maps. Differentiation between non-atypical and malignant (GP ≥ 3) areas resulted in an accuracy of 92% with a sensitivity and specificity of 90 and 93%, respectively. The differentiation between GP ≥ 4 and GP ≤ 3 was accurate for 90%, with a sensitivity and specificity of 77 and 94%, respectively . Concordance of our automated GG determination method with a genitourinary pathologist was obtained in 65% ( κ = 0.70), indicating substantial agreement. A CNN allows for accurate differentiation between non-atypical and malignant areas as defined by GPs, leading to a substantial agreement with the pathologist in defining the GG. Electronic supplementary material The online version of this article (10.1007/s00428-019-02577-x) contains supplementary material, which is available to authorized users.