Infi ltration of human melanomas with cytotoxic immune cells correlates with spontaneous type I IFN activation and a favorable prognosis. Therapeutic blockade of immune-inhibitory receptors in patients with preexisting lymphocytic infi ltrates prolongs survival, but new complementary strategies are needed to activate cellular antitumor immunity in immune cell-poor melanomas. Here, we show that primary melanomas in Hgf-Cdk4 R24C mice, which imitate human immune cell-poor melanomas with a poor outcome, escape IFN-induced immune surveillance and editing. Peritumoral injections of immunostimulatory RNA initiated a cytotoxic infl ammatory response in the tumor microenvironment and signifi cantly impaired tumor growth. This critically required the coordinated induction of type I IFN responses by dendritic, myeloid, natural killer, and T cells. Importantly, antibodymediated blockade of the IFN-induced immune-inhibitory interaction between PD-L1 and PD-1 receptors further prolonged the survival. These results highlight important interconnections between type I IFNs and immune-inhibitory receptors in melanoma pathogenesis, which serve as targets for combination immunotherapies. SIGNIFICANCE:Using a genetically engineered mouse melanoma model, we demonstrate that targeted activation of the type I IFN system with immunostimulatory RNA in combination with blockade of immune-inhibitory receptors is a rational strategy to expose immune cell-poor tumors to cellular immune surveillance. Cancer Discov; 4(6);

Background: Recent studies have successfully demonstrated the use of deep-learning algorithms for dermatologist-level classification of suspicious lesions by the use of excessive proprietary image databases and limited numbers of dermatologists. For the first time, the performance of a deep-learning algorithm trained by open-source images exclusively is compared to a large number of dermatologists covering all levels within the clinical hierarchy. Methods: We used methods from enhanced deep learning to train a convolutional neural network (CNN) with 12,378 open-source dermoscopic images. We used 100 images to compare the performance of the CNN to that of the 157 dermatologists from 12 university hospitals in Germany.