Explore the vast range of titles available.

For virtually every patient with colorectal cancer (CRC), hematoxylin-eosin (HE)-stained tissue slides are available. These images contain quantitative information, which is not routinely used to objectively extract prognostic biomarkers. In the present study, we investigated whether deep convolutional neural networks (CNNs) can extract prognosticators directly from these widely available images. We hand-delineated single-tissue regions in 86 CRC tissue slides, yielding more than 100,000 HE image patches, and used these to train a CNN by transfer learning, reaching a nine-class accuracy of >94% in an independent data set of 7,180 images from 25 CRC patients. With this tool, we performed automated tissue decomposition of representative multitissue HE images from 862 HE slides in 500 stage I-IV CRC patients in the The Cancer Genome Atlas (TCGA) cohort, a large international multicenter collection of CRC tissue. Based on the output neuron activations in the CNN, we calculated a \"deep stroma score,\" which was an independent prognostic factor for overall survival (OS) in a multivariable Cox proportional hazard model (hazard ratio [HR] with 95% confidence interval [CI]: 1.99 [1.27-3.12], p = 0.0028), while in the same cohort, manual quantification of stromal areas and a gene expression signature of cancer-associated fibroblasts (CAFs) were only prognostic in specific tumor stages. We validated these findings in an independent cohort of 409 stage I-IV CRC patients from the \"Darmkrebs: Chancen der Verhütung durch Screening\" (DACHS) study who were recruited between 2003 and 2007 in multiple institutions in Germany. Again, the score was an independent prognostic factor for OS (HR 1.63 [1.14-2.33], p = 0.008), CRC-specific OS (HR 2.29 [1.5-3.48], p = 0.0004), and relapse-free survival (RFS; HR 1.92 [1.34-2.76], p = 0.0004). A prospective validation is required before this biomarker can be implemented in clinical workflows. In our retrospective study, we show that a CNN can assess the human tumor microenvironment and predict prognosis directly from histopathological images.

Lymphoid and myeloid cells are abundant in the tumor microenvironment, can be quantified by immunohistochemistry and shape the disease course of human solid tumors. Yet, there is no comprehensive understanding of spatial immune infiltration patterns (‘topography’) across cancer entities and across various immune cell types. In this study, we systematically measure the topography of multiple immune cell types in 965 histological tissue slides from N = 177 patients in a pan-cancer cohort. We provide a definition of inflamed (‘hot’), non-inflamed (‘cold’) and immune excluded patterns and investigate how these patterns differ between immune cell types and between cancer types. In an independent cohort of N = 287 colorectal cancer patients, we show that hot, cold and excluded topographies for effector lymphocytes (CD8) and tumor-associated macrophages (CD163) alone are not prognostic, but that a bivariate classification system can stratify patients. Our study adds evidence to consider immune topographies as biomarkers for patients with solid tumors.

Automatic recognition of different tissue types in histological images is an essential part in the digital pathology toolbox. Texture analysis is commonly used to address this problem; mainly in the context of estimating the tumour/stroma ratio on histological samples. However, although histological images typically contain more than two tissue types, only few studies have addressed the multi-class problem. For colorectal cancer, one of the most prevalent tumour types, there are in fact no published results on multiclass texture separation. In this paper we present a new dataset of 5,000 histological images of human colorectal cancer including eight different types of tissue. We used this set to assess the classification performance of a wide range of texture descriptors and classifiers. As a result, we found an optimal classification strategy that markedly outperformed traditional methods, improving the state of the art for tumour-stroma separation from 96.9% to 98.6% accuracy and setting a new standard for multiclass tissue separation (87.4% accuracy for eight classes). We make our dataset of histological images publicly available under a Creative Commons license and encourage other researchers to use it as a benchmark for their studies.

Patients with ulcerative colitis (UC) are at increased risk for developing colorectal cancer (CRC). In contrast to sporadic colorectal tumorigenesis, TP53 mutations occur early in the progression from inflamed colonic epithelium to dysplasia to CRC, and are sometimes readily detectable in inflamed, (yet) non-dysplastic mucosa. Here, we analyzed formalin-fixed paraffin-embedded tissue samples from 19 patients with long-standing UC (median 18 years, range 3 to 34) who had developed CRC as a consequence of chronic inflammation of the large bowel. We performed microsatellite instability testing, copy number analysis by array-based comparative genomic hybridization, mutation analysis by targeted next generation sequencing (48-gene panel) and TP53 immunostaining. The results were compared to The Cancer Genome Atlas (TCGA) data on sporadic CRC. All UC-CRC lesions in our cohort were microsatellite stable. Overall, genomic imbalances of UC-CRCs showed patterns of chromosomal aneuploidies characteristic for sporadic CRC with the exception of gains of chromosome arm 5p (12 of 23 UC-CRC, 52%), which are rare in sporadic CRCs from TCGA (21 of 144, 15%; FDR adjusted P = 0.006). UC-CRCs showed a predilection for TP53 alterations, which was the most frequently mutated gene in our cohort (20 of 23, 87%). Interestingly, spatially separated tumor lesions from individual patients tended to harbor distinct TP53 mutations. Similar to CRCs arising in a background of Crohn's colitis, the genetic landscape of UC-CRCs was characterized by TP53 mutations and chromosomal aneuploidies including gains of chromosome arm 5p. Both alterations harbor the potential for early detection in precursor lesions, thus complementing morphologic diagnosis.

Patient-derived organoids resemble the biology of tissues and tumors, enabling ex vivo modeling of human diseases. They have heterogeneous morphologies with unclear biological causes and relationship to treatment response. Here, we use high-throughput, image-based profiling to quantify phenotypes of over 5 million individual colorectal cancer organoids after treatment with >500 small molecules. Integration of data using multi-omics modeling identifies axes of morphological variation across organoids: Organoid size is linked to IGF1 receptor signaling, and cystic vs. solid organoid architecture is associated with LGR5 + stemness. Treatment-induced organoid morphology reflects organoid viability, drug mechanism of action, and is biologically interpretable. Inhibition of MEK leads to cystic reorganization of organoids and increases expression of LGR5 , while inhibition of mTOR induces IGF1 receptor signaling. In conclusion, we identify shared axes of variation for colorectal cancer organoid morphology, their underlying biological mechanisms, and pharmacological interventions with the ability to move organoids along them. The heterogeneity underlying cancer organoid phenotypes is not yet well understood. Here, the authors develop an imaging analysis assay for high throughput phenotypic screening of colorectal organoids that allows to define specific morphological changes that occur following different drug treatments.

Background HER2 is considered as one of the most important, predictive biomarkers in oncology. The diagnosis of HER2 positive cancer types such as breast- and gastric cancer is usually based on immunohistochemical HER2 staining of tumour tissue. However, the current immunohistochemical methods do not provide localized information about HER2’s functional state. In order to generate signals leading to cell growth and proliferation, the receptor spontaneously forms homodimers, a process that can differ between individual cancer cells. Materials and methods HER2 overexpressing tumour cells were dissociated from formalin-fixed paraffin-embedded (FFPE) patient’s biopsy sections, subjected to a heat-induced antigen retrieval procedure, and immobilized on microchips. HER2 was specifically labelled via a two-step protocol involving the incubation with an Affibody-biotin compound followed by the binding of a streptavidin coated quantum dot (QD) nanoparticle. Cells with membrane bound HER2 were identified using fluorescence microscopy, coated with graphene to preserve their hydrated state, and subsequently examined by scanning transmission electron microscopy (STEM) to obtain the locations at the single molecule level. Label position data was statistically analysed via the pair correlation function, yielding information about the presence of HER2 homodimers. Results Tumour cells from two biopsies, scored HER2 3+, and a HER2 negative control sample were examined. The specific labelling protocol was first tested for a sectioned tissue sample of HER2-overexpressing tumour. Subsequently, a protocol was optimized to study HER2 homodimerization in single cells dissociated from the tissue section. Electron microscopy data showed membrane bound HER2 in average densities of 201–689 proteins/μm 2 . An automated, statistical analysis of well over 200,000 of measured protein positions revealed the presence of HER2 homodimers in 33 and 55% of the analysed images for patient 1 and 2, respectively. Conclusions We introduced an electron microscopy method capable of measuring the positions of individually labelled HER2 proteins in patient tumour cells from which information about the functional status of the receptor was derived. This method could take HER2 testing a step further by examining HER2 homodimerization directly out of tumour tissue and may become important for adjusting a personalized antibody-based drug therapy.

For clear cell renal cell carcinoma (ccRCC) risk-dependent diagnostic and therapeutic algorithms are routinely implemented in clinical practice. Artificial intelligence-based image analysis has the potential to improve outcome prediction and thereby risk stratification. Thus, we investigated whether a convolutional neural network (CNN) can extract relevant image features from a representative hematoxylin and eosin-stained slide to predict 5-year overall survival (5y-OS) in ccRCC. The CNN was trained to predict 5y-OS in a binary manner using slides from TCGA and validated using an independent in-house cohort. Multivariable logistic regression was used to combine of the CNNs prediction and clinicopathological parameters. A mean balanced accuracy of 72.0% (standard deviation [SD] = 7.9%), sensitivity of 72.4% (SD = 10.6%), specificity of 71.7% (SD = 11.9%) and area under receiver operating characteristics curve (AUROC) of 0.75 (SD = 0.07) was achieved on the TCGA training set (n = 254 patients / WSIs) using 10-fold cross-validation. On the external validation cohort (n = 99 patients / WSIs), mean accuracy, sensitivity, specificity and AUROC were 65.5% (95%-confidence interval [CI]: 62.9–68.1%), 86.2% (95%-CI: 81.8–90.5%), 44.9% (95%-CI: 40.2–49.6%), and 0.70 (95%-CI: 0.69–0.71). A multivariable model including age, tumor stage and metastasis yielded an AUROC of 0.75 on the TCGA cohort. The inclusion of the CNN-based classification (Odds ratio = 4.86, 95%-CI: 2.70–8.75, p < 0.01) raised the AUROC to 0.81. On the validation cohort, both models showed an AUROC of 0.88. In univariable Cox regression, the CNN showed a hazard ratio of 3.69 (95%-CI: 2.60–5.23, p < 0.01) on TCGA and 2.13 (95%-CI: 0.92–4.94, p = 0.08) on external validation. The results demonstrate that the CNN’s image-based prediction of survival is promising and thus this widely applicable technique should be further investigated with the aim of improving existing risk stratification in ccRCC.

Inositol-requiring enzyme 1 (IRE1) is a proximal endoplasmic reticulum (ER) stress sensor and a central mediator of the unfolded protein response. In a human glioma model, inhibition of IRE1α correlated with down-regulation of prevalent proangiogenic factors such as VEGF-A, IL-1β, IL-6, and IL-8. Significant up-regulation of antiangiogenic gene transcripts was also apparent. These transcripts encode SPARC, decorin, thrombospondin-1, and other matrix proteins functionally linked to mesenchymal differentiation and glioma invasiveness. In vivo, using both the chick chorio-allantoic membrane assay and a mouse orthotopic brain model, we observed in tumors underexpressing IRE1: (i) reduction of angiogenesis and blood perfusion, (ii) a decreased growth rate, and (iii) extensive invasiveness and blood vessel cooption. This phenotypic change was consistently associated with increased overall survival in glioma-implanted recipient mice. Ectopic expression of IL-6 in IRE1-deficient tumors restored angiogenesis and neutralized vessel cooption but did not reverse the mesenchymal/infiltrative cell phenotype. The ischemia-responsive IRE1 protein is thus identified as a key regulator of tumor neovascularization and invasiveness.

Human cancer cell lines are frequently used as model systems to study molecular mechanisms and genetic changes in cancer. However, the model is repeatedly criticized for its lack of proximity to original patient tumors. Therefore, understanding to what extent cell lines cultured under artificial conditions reflect the phenotypic and genomic profiles of their corresponding parental tumors is crucial when analyzing their biological properties. To directly compare molecular alterations between patient tumors and derived cell lines, we have established new cancer cell lines from four patients with gastrointestinal tumors. Tumor entities comprised esophageal cancer, colon cancer, rectal cancer and pancreatic cancer. Phenotype and genotype of both patient tumors and derived low-passage cell lines were characterized by immunohistochemistry (22 different antibodies), array-based comparative genomic hybridization and targeted next generation sequencing (48-gene panel). The immunophenotype was highly consistent between patient tumors and derived cell lines; the expression of most markers in cell lines was concordant with the respective parental tumor and characteristic for the respective tumor entities in general. The chromosomal aberration patterns of the parental tumors were largely maintained in the cell lines and the distribution of gains and losses was typical for the respective cancer entity, despite a few distinct differences. Cancer gene mutations (e.g., KRAS , TP53 ) and microsatellite status were also preserved in the respective cell line derivates. In conclusion, the four examined newly established cell lines exhibited a phenotype and genotype closely recapitulating their parental tumor. Hence, newly established cancer cell lines may be useful models for further pharmacogenomic studies.

Renal cell carcinoma (RCC) is the deadliest primary genitourinary malignancy typically associated with asymptomatic initial presentation and poorly predictable survival. Next to established risk factors, tumor microenvironment may alter metastatic capacity and immune landscape. Due to their high concentrations, sulfoglycolipids (sulfatides) were among the first well-described antigens in RCC that are associated with worse prognosis. As sulfatide detection in routine diagnostics is not possible, we aimed to test the prognostic value of its protein counterpart, sulfatide-producing enzyme Gal3ST1. We performed retrospective long-term follow up analysis of Gal3ST1 expression as prognostic risk factor in a representative RCC patient cohort. We observed differentially regulated Gal3ST1 expression in all RCC types, being significantly more associated with clear cell RCC than to chromophobe RCC (p = 0.001). Surprisingly, in contrast to published observations from in vitro models, we could not confirm an association between Gal3ST1 expression and a malignant clinical behaviour of the RCC. In our cohort, Gal3ST1 did not significantly influence progression-free survival (Hazard Ratio (HR): 1.7 95% CI (0.6–4.9), p = 0.327). Particularly after adjusting for histology, T-stage, N-status and M-status at baseline, we observed no independent prognostic effect (HR = 1.0 95% CI (0.3–3.3), p = 0.96). The analysis of Gal3ST1 mRNA expression in a TCGA dataset supported the results of our cohort. Thus, Gal3ST1 might help to differentiate between chromophobe RCC and other frequent RCC entities but—despite previously published data from cell culture models—does not qualify as a prognostic marker for RCC. Further investigation of regulatory mechanisms of sulfatide metabolism in human RCC microenvironment is necessary to understand the role of this quantitatively prominent glycosphingolipid in RCC progression.