Search

Contains fulltext : 293170.pdf (Publisher’s version ) (Open Access), In colorectal cancer (CRC), artificial intelligence (AI) can alleviate the laborious task of characterization and reporting on resected biopsies, including polyps, the numbers of which are increasing as a result of CRC population screening programs ongoing in many countries all around the globe. Here, we present an approach to address two major challenges in the automated assessment of CRC histopathology whole-slide images. We present an AI-based method to segment multiple ([Formula: see text]) tissue compartments in the H &E-stained whole-slide image, which provides a different, more perceptible picture of tissue morphology and composition. We test and compare a panel of state-of-the-art loss functions available for segmentation models, and provide indications about their use in histopathology image segmentation, based on the analysis of (a) a multi-centric cohort of CRC cases from five medical centers in the Netherlands and Germany, and (b) two publicly available datasets on segmentation in CRC. We used the best performing AI model as the basis for a computer-aided diagnosis system that classifies colon biopsies into four main categories that are relevant pathologically. We report the performance of this system on an independent cohort of more than 1000 patients. The results show that with a good segmentation network as a base, a tool can be developed which can support pathologists in the risk stratification of colorectal cancer patients, among other possible uses. We have made the segmentation model available for research use on https://grand-challenge.org/algorithms/colon-tissue-segmentation/ .

Background: Surveillance of individuals at risk of developing pancreatic ductal adenocarcinoma (PDAC) has the potential to improve survival, yet early detection based on solely imaging modalities is challenging. We aimed to identify changes in serum glycosylation levels over time to earlier detect PDAC in high-risk individuals. Methods: Individuals with a hereditary predisposition to develop PDAC were followed in two surveillance programs. Those, of which at least two consecutive serum samples were available, were included. Mass spectrometry analysis was performed to determine the total N-glycome for each consecutive sample. Potentially discriminating N-glycans were selected based on our previous cross-sectional analysis and relative abundances were calculated for each glycosylation feature. Results: 165 individuals (“FPC-cohort” N = 119; Leiden cohort N = 46) were included. In total, 97 (59%) individuals had a genetic predisposition (77 CDKN2A, 15 BRCA1/2, 5 STK11) and 68 (41%) a family history of PDAC without a known genetic predisposition (>10-fold increased risk of developing PDAC). From each individual, a median number of 3 serum samples (IQR 3) was collected. Ten individuals (6%) developed PDAC during 35 months of follow-up; nine (90%) of these patients carried a CDKN2A germline mutation. In PDAC cases, compared to all controls, glycosylation characteristics were increased (fucosylation, tri- and tetra-antennary structures, specific sialic linkage types), others decreased (complex-type diantennary and bisected glycans). The largest change over time was observed for tri-antennary fucosylated glycans, which were able to differentiate cases from controls with a specificity of 92%, sensitivity of 49% and accuracy of 90%. Conclusion: Serum N-glycan monitoring may support early detection in a pancreas surveillance program.

Introduction: The tumor-associated stoma is of importance for tumor progression and is generally accepted to have a significant influence on patient prognosis. However, little is known regarding specific features of tumor-associated stromal tissues and response to (neoadjuvant) chemotherapy. This study investigated the predictive value of extracellular matrix organization on response to chemotherapy in patients treated in the NEOZOTAC trial. Methods: Stromal organisation was analyzed via a simple method using image analysis software on hematoxylin and eosin (H&E)-stained slides from primary tumor biopsies collected as part of the NEOZOTAC trial. Heidenhain's AZAN trichrome-stained slides were also analyzed for comparison of collagen evaluation. Sections were stained for phospho-Smad2 (pS2) in order to determine the relationship of TGF-beta signaling with stromal organization. Results: A statistically significant relationship was observed between stroma consisting of organised collagen and pathological response to neoadjuvant chemotherapy (Odds Ratio 0.276, 95%CI 0.124-0.614, P = 0.002). This parameter was also related to ER-status (P = 0.003), clinical tumor -status (P = 0.041), nodal status (P = 0.029) and pS2 status (P = 0.025). Correlation between stromal organisation determined on H&E-stained and AZAN-stained tissue sections was high (Pearson's correlation coefficient = 0.806). Conclusion: Intratumoral stromal organisation determined using pre-treatment breast cancer biopsies was related to pathological response to chemotherapy. This parameter might play a role in the management of breast cancer for identifying those patients that are likely to benefit from neoadjuvant chemotherapy.