Microsatellite instability (MSI) has been approved as a pan-cancer biomarker for immune checkpoint blockade (ICB) therapy. However, current MSI identification methods are not available for all patients. We proposed an ensemble multiple instance deep learning model to predict microsatellite status based on histopathology images, and interpreted the pathomics-based model with multi-omics correlation. Methods: Two cohorts of patients were collected, including 429 from The Cancer Genome Atlas (TCGA-COAD) and 785 from an Asian colorectal cancer (CRC) cohort (Asian-CRC). We established the pathomics model, named Ensembled Patch Likelihood Aggregation (EPLA), based on two consecutive stages: patch-level prediction and WSI-level prediction. The initial model was developed and validated in TCGA-COAD, and then generalized in Asian-CRC through transfer learning. The pathological signatures extracted from the model were analyzed with genomic and transcriptomic profiles for model interpretation. Results: The EPLA model achieved an area-under-the-curve (AUC) of 0.8848 (95% CI: 0.8185-0.9512) in the TCGA-COAD test set and an AUC of 0.8504 (95% CI: 0.7591-0.9323) in the external validation set Asian-CRC after transfer learning. Notably, EPLA captured the relationship between pathological phenotype of poor differentiation and MSI ( P < 0.001). Furthermore, the five pathological imaging signatures identified from the EPLA model were associated with mutation burden and DNA damage repair related genotype in the genomic profiles, and antitumor immunity activated pathway in the transcriptomic profiles. Conclusions: Our pathomics-based deep learning model can effectively predict MSI from histopathology images and is transferable to a new patient cohort. The interpretability of our model by association with pathological, genomic and transcriptomic phenotypes lays the foundation for prospective clinical trials of the application of this artificial intelligence (AI) platform in ICB therapy.

Keloid is a skin fibrotic disease with the characteristics of recurrence and invasion, its pathogenesis still remains unrevealed. The epithelial-mesenchymal transition (EMT) is critical for wound healing, fibrosis, recurrence, and invasion of cancer. We sought to investigate the EMT in keloid and the mechanism through which the EMT regulates keloid formation. In keloid tissues, the expressions of EMT-associated markers and transforming growth factor (TGF)-β1/Smad3 signaling were examined by immunohistochemistry. In the keloid epidermis and dermal tissue, the expressions of genes related to the regulation of skin homeostasis, fibroblast growth factor receptor 2 (FGFR2) and p63, were analyzed using quantitative real-time polymerase chain reaction. The results showed that accompanying the loss of the epithelial marker E-cadherin and the gain of the mesenchymal markers fibroblast-specific protein 1 (FSP1) and vimentin in epithelial cells from epidermis and skin appendages, and in endothelial cells from dermal microvessels, enhanced TGF-β1 expression and Smad3 phosphorylation were noted in keloid tissues. Moreover, alternative splicing of the FGFR2 gene switched the predominantly expressed isoform from FGFR2-IIIb to -IIIc, concomitant with the decreased expression of ΔNp63 and TAp63, which changes might partially account for abnormal epidermis and appendages in keloids. In addition, we found that TGF-β1-induced hair follicle outer root sheath keratinocytes (ORSKs) and normal skin epithelial cells underwent EMT in vitro with ORSKs exhibiting more obvious EMT changes and more similar expression profiles for EMT-associated and skin homeostasis-related genes as in keloid tissues, suggesting that ORSKs might play crucial roles in the EMT in keloids. Our study provided insights into the molecular mechanisms mediating the EMT pathogenesis of keloids.

Non-small cell lung cancer (NSCLC) is frequently associated with oncogenic driver mutations, which play an important role in carcinogenesis and cancer progression. Targeting epidermal growth factor receptor (EGFR) mutations and anaplastic lymphoma kinase rearrangements has become standard therapy for patients with these aberrations because of the greater improvement of survival, tolerance, and quality-of-life compared to chemotherapy. Clinical trials for emerging therapies that target other less common driver genes are generating mixed results. Here, we review the literature on rare drivers in NSCLC with frequencies lower than 5% (e.g., ROS1, RET, MET, BRAF, NTRK, HER2, NRG1, FGFR1, PIK3CA, DDR2, and EGFR exon 20 insertions). In summary, targeting rare oncogenic drivers in NSCLC has achieved some success. With the development of new inhibitors that target these rare drivers, the spectrum of targeted therapy has been expanded, although acquired resistance is still an unavoidable problem.

Background: This research aims to investigate the evaluation methods of teaching oral implant clinical courses and estimate the effectiveness of a virtual simulation platform. Methods: Eighty second-and third-year undergraduates in Lanzhou University were recruited and randomized to either three experimental groups or one control group. The subjects undertook theoretical examinations to test their basic level of knowledge after training in similarly unified knowledge courses. Each student group then participated in an eight-hour operating training session. An operation test on pig mandible was conducted, followed by a second theoretical examination. The assessment consists of three distinct parts: a subjective operating score by a clinical senior teacher, an implant accuracy analysis in cone-beam computed tomography (angular, apical, and entrance deviation), and comparison of the two theoretical examinations. Finally, students completed a questionnaire gauging their understanding of the virtual simulation. Results: There was no significant difference between the four groups in first theoretical examination (P > 0.05); the second theoretical scores of the V-J and J-V group (62.90 ± 3.70, 60.05 ± 2.73) were significantly higher than the first time (57.05 ± 3.92, P < 0.05), while no difference between the V (57.10 ± 3.66) and J (56.89 ± 2.67) groups was found. Thus, the combination of V-J was effective in improving students' theoretical scores. The V-J and J-V groups had higher scores on operation (73.98 ± 4.58, 71.85 ± 4.67) and showed better implant precision. Conclusion: Virtual simulation education, especially with a jaw simulation model, could improve students' implantology achievements and training. Currently study found that the V-J group may performed better than the J-V group in oral implant teaching.