Comparing deep learning and pathologist quantification of cell-level PD-L1 expression in non-small cell lung cancer whole-slide images.

Programmed death-ligand 1 (PD-L1) expression is currently used in the clinic to assess eligibility for immune-checkpoint inhibitors via the tumor proportion score (TPS), but its efficacy is limited by high interobserver variability. Multiple papers have presented systems for the automatic quantification of TPS, but none report on the task of determining cell-level PD-L1 expression and often reserve their evaluation to a single PD-L1 monoclonal antibody or clinical center. In this paper, we report on a deep learning algorithm for detecting PD-L1 negative and positive tumor cells at a cellular level and evaluate it on a cell-level reference standard established by six readers on a multi-centric, multi PD-L1 assay dataset. This reference standard also provides for the first time a benchmark for computer vision algorithms. In addition, in line with other papers, we also evaluate our algorithm at slide-level by measuring the agreement between the algorithm and six pathologists on TPS quantification. We find a moderately low interobserver agreement at cell-level level (mean reader-reader F1 score = 0.68) which our algorithm sits slightly under (mean reader-AI F1 score = 0.55), especially for cases from the clinical center not included in the training set. Despite this, we find good AI-pathologist agreement on quantifying TPS compared to the interobserver agreement (mean reader-reader Cohen's kappa = 0.54, 95% CI 0.26-0.81, mean reader-AI kappa = 0.49, 95% CI 0.27-0.72). In conclusion, our deep learning algorithm demonstrates promise in detecting PD-L1 expression at a cellular level and exhibits favorable agreement with pathologists in quantifying the tumor proportion score (TPS). We publicly release our models for use via the Grand-Challenge platform.

Development of a miRNA-based classifier for detection of colorectal cancer molecular subtypes.

Previously, colorectal cancer (CRC) has been classified into four distinct molecular subtypes based on transcriptome data. These consensus molecular subtypes (CMSs) have implications for our understanding of tumor heterogeneity and the prognosis of patients. So far, this classification has been based on the use of messenger RNAs (mRNAs), although microRNAs (miRNAs) have also been shown to play a role in tumor heterogeneity and biological differences between CMSs. In contrast to mRNAs, miRNAs have a smaller size and increased stability, facilitating their detection. Therefore, we built a miRNA-based CMS classifier by converting the existing mRNA-based CMS classification using machine learning (training dataset of n = 271). The performance of this miRNA-assigned CMS classifier (CMS-miRaCl) was evaluated in several datasets, achieving an overall accuracy of ~ 0.72 (0.6329-0.7987) in the largest dataset (n = 158). To gain insight into the biological relevance of CMS-miRaCl, we evaluated the most important features in the classifier. We found that miRNAs previously reported to be relevant in microsatellite-instable CRCs or Wnt signaling were important features for CMS-miRaCl. Following further studies to validate its robustness, this miRNA-based alternative might simplify the implementation of CMS classification in clinical workflows.

Enhanced Collagen Deposition in the Duodenum of Patients with Hyaline Fibromatosis Syndrome and Protein Losing Enteropathy.

Hyaline fibromatosis syndrome (HFS), resulting from ANTXR2 mutations, is an ultra-rare disease that causes intestinal lymphangiectasia and protein-losing enteropathy (PLE). The mechanisms leading to the gastrointestinal phenotype in these patients are not well defined. We present two patients with congenital diarrhea, severe PLE and unique clinical features resulting from deleterious ANTXR2 mutations. Intestinal organoids were generated from one of the patients, along with CRISPR-Cas9 ANTXR2 knockout, and compared with organoids from two healthy controls. The ANTXR2-deficient organoids displayed normal growth and polarity, compared to controls. Using an anthrax-toxin assay we showed that the c.155C>T mutation causes loss-of-function of ANTXR2 protein. An intrinsic defect of monolayer formation in patient-derived or ANTXR2KO organoids was not apparent, suggesting normal epithelial function. However, electron microscopy and second harmonic generation imaging showed abnormal collagen deposition in duodenal samples of these patients. Specifically, collagen VI, which is known to bind ANTXR2, was highly expressed in the duodenum of these patients. In conclusion, despite resistance to anthrax-toxin, epithelial cell function, and specifically monolayer formation, is intact in patients with HFS. Nevertheless, loss of ANTXR2-mediated signaling leads to collagen VI accumulation in the duodenum and abnormal extracellular matrix composition, which likely plays a role in development of PLE.