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.

Thymomas and thymic carcinomas (TCs) are epithelial tumors of the thymus. They are the most frequent tumors of the anterior mediastinum in adults but account for less than 1% of all human neoplasms. Thymomas comprise a spectrum of unique tumors of low to moderate malignant potential that are subdivided according to the histological classification of the World Health Organization (WHO) into type A, AB, B1, B3, and B3 and rare other subtypes. By contrast, TCs resemble similarly termed carcinomas outside the mediastinum and usually follow an aggressive clinical course.¹ x 1 Travis, W, Brambilla, E, Müller-Hermelink, H et al. Pathology and Genetics of Tumours of the Lung, Pleura, Thymus and Heart. IARC Press, Lyon; 2004 Google Scholar See all References

Pancreatic ductal adenocarcinoma (PDAC) is frequently accompanied by excruciating pain, which has been associated with attraction of cancer cells and their invasion of intrapancreatic sensory nerves. Neutralization of the chemokine CCL2 reduced cancer-associated pain in a clinical trial, but there have been no systematic analyses of the highly diverse chemokine families and their receptors in PDAC. We performed an open, unbiased RNA-interference screen of mammalian chemokines in co-cultures of mouse PDAC cells (K8484) and mouse peripheral sensory neurons, and confirmed findings in studies of DT8082 PDAC cells. We studied the effects of chemokines on migration of PDAC cell lines. Orthotopic tumors were grown from K8484 cells in mice, and mice were given injections of neutralizing antibodies against chemokines, antagonists, or control antibodies. We analyzed abdominal mechanical hypersensitivity and collected tumors and analyzed them by histology and immunohistochemistry to assess neural remodeling. We collected PDAC samples and information on pain levels from 74 patients undergoing resection and measured levels of CXCR3 and CCR7 by immunohistochemistry and immunoblotting. Knockdown of 9 chemokines in DRG neurons significantly reduced migration of PDAC cells towards sensory neurons. Sensory neuron-derived CCL21 and CXCL10 promoted migration of PDAC cells via their receptors CCR7 and CXCR3, respectively, which were expressed by cells in orthotopic tumors and PDAC specimens from patients. Neutralization of CCL21 or CXCL10, or their receptors, in mice with orthotopic tumors significantly reduced nociceptive hypersensitivity and nerve fiber hypertrophy and improved behavioral parameters without affecting tumor infiltration by T cells or neutrophils. Increased levels of CXCR3 and CCR7 in human PDAC specimens were associated with increased frequency of cancer-associated pain, determined from patient questionnaires. In an unbiased screen of chemokines, we identified CCL21 and CXCL10 as proteins that promote migration of pancreatic cancer cells toward sensory neurons. Inhibition of these chemokines or their receptors reduce hypersensitivity in mice with orthotopic tumors, and patients with PDACs with high levels of the chemokine receptors of CXCR3 and CCR7 had increased frequency of cancer-associated pain.

Reliable entity subtyping is paramount for therapy stratification in lung cancer. Morphological evaluation remains the basis for entity subtyping and directs the application of additional methods such as immunohistochemistry (IHC). The decision of whether to perform IHC for subtyping is subjective, and access to IHC is not available worldwide. Thus, the application of additional methods to support morphological entity subtyping is desirable. Therefore, the ability of convolutional neuronal networks (CNNs) to classify the most common lung cancer subtypes, pulmonary adenocarcinoma (ADC), pulmonary squamous cell carcinoma (SqCC), and small-cell lung cancer (SCLC), was evaluated. A cohort of 80 ADC, 80 SqCC, 80 SCLC, and 30 skeletal muscle specimens was assembled; slides were scanned; tumor areas were annotated; image patches were extracted; and cases were randomly assigned to a training, validation or test set. Multiple CNN architectures (VGG16, InceptionV3, and InceptionResNetV2) were trained and optimized to classify the four entities. A quality control (QC) metric was established. An optimized InceptionV3 CNN architecture yielded the highest classification accuracy and was used for the classification of the test set. Image patch and patient-based CNN classification results were 95% and 100% in the test set after the application of strict QC. Misclassified cases mainly included ADC and SqCC. The QC metric identified cases that needed further IHC for definite entity subtyping. The study highlights the potential and limitations of CNN image classification models for tumor differentiation.

Background: Histological images show strong variance (e.g. illumination, color, staining quality) due to differences in image acquisition, tissue processing, staining, etc. This can impede downstream image analysis such as staining intensity evaluation or classification. Methods to reduce these variances are called image normalization techniques. Methods: In this paper, we investigate the potential of CycleGAN (cycle consistent Generative Adversarial Network) for color normalization in hematoxylin-eosin stained histological images using daily clinical data with consideration of the variability of internal staining protocol variations. The network consists of a generator network G_B that learns to map an image X from a source domain A to a target domain B, i.e. G_B:X_A→X_B. In addition, a discriminator network D_B is trained to distinguish whether an image from domain B is real or generated. The same process is applied to another generator-discriminator pair (G_A,D_A), for the inverse mapping G_A:X_B→X_A. Cycle consistency ensures that a generated image is close to its original when being mapped backwards (G_A(G_B(X_A))≈X_A and vice versa). We validate the CycleGAN approach on a breast cancer challenge and a follicular thyroid carcinoma data set for various stain variations. We evaluate the quality of the generated images compared to the original images using similarity measures. In addition, we apply stain normalization on pathological lymph node data from our institute and test the gain from normalization on a ResNet classifier pre-trained on the Camelyon16 data set. Results: Qualitative results of the images generated by our network are compared to original color distributions. Our evaluation indicates that by mapping images to a target domain, the similarity training images from that domain improves up to 96%. We also achieve a high cycle consistency for the generator networks by obtaining similarity indices greater than 0.9. When applying the CycleGAN normalization to HE-stain images from our institute the kappa-value of the ResNet-model that is only trained on Camelyon16 data is increased more than 50%. Conclusions: CycleGANs have proven to efficiently normalize HE-stained images. The approach compensates for deviations resulting from image acquisition (e.g. different scanning devices) as well as from tissue staining (e.g. different staining protocols), and thus overcomes the staining variations in images from various institutions.The code is publicly available at https://github.com/m4ln/stainTransfer_CycleGAN_pytorch. The data set supporting the solutions is available at 10.11588/data/8LKEZF.