Colonoscopy is commonly used to screen for colorectal cancer (CRC). We develop a deep learning model called CRCNet for optical diagnosis of CRC by training on 464,105 images from 12,179 patients and test its performance on 2263 patients from three independent datasets. At the patient-level, CRCNet achieves an area under the precision-recall curve (AUPRC) of 0.882 (95% CI: 0.828–0.931), 0.874 (0.820–0.926) and 0.867 (0.795–0.923). CRCNet exceeds average endoscopists performance on recall rate across two test sets (91.3% versus 83.8%; two-sided t-test, p < 0.001 and 96.5% versus 90.3%; p = 0.006) and precision for one test set (93.7% versus 83.8%; p = 0.02), while obtains comparable recall rate on one test set and precision on the other two. At the image-level, CRCNet achieves an AUPRC of 0.990 (0.987–0.993), 0.991 (0.987–0.995), and 0.997 (0.995–0.999). Our study warrants further investigation of CRCNet by prospective clinical trials.

Background: Lymph node metastasis (LNM) occurs frequently in young papillary thyroid carcinoma (PTC) patients, though the mortality rates are low. We aimed to analyze the relationship between age at diagnosis and LNM in PTC at a population level to elucidate the clinical behavior of PTC. Methods: Data of adult patients with surgically treated PTC and follicular thyroid carcinoma (FTC) were identified from the Surveillance, Epidemiology, and End Results (SEER) database (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) to investigate the relationship between age and clinical characteristics by curve estimation. The adjusted odds ratio of age and LNM rate were determined.Results: A total of 50,347 PTC (48,166) and FTC (2181) (median age: 45 and 50 years, respectively) patients met the inclusion criteria; 44.5% of those with PTC (21,428) had LNM. Rank-sum test analysis indicated differences in distribution of age in LNM-positive and LNM-negative PTC. The relationship between age, tumor size and LNM showed a quadratic curve in PTC. The mean tumor diameter and LNM rate correlated linearly with age in 18-59year-old patients. LNM rate decreased with age (R 2 = 0.932, P < .0001), especially women (R 2 = 0.951, P < .0001). Conclusion:In young and middle-aged PTC patients, LNM may resolve spontaneously with delayed diagnosis and management. Active surveillance of low-risk PTC is justified.

ObjectiveThe objective of this research was to screen prognostic related genes of thyroid papillary carcinoma (PTC) by single-cell RNA sequencing (scRNA-seq), to construct the diagnostic and prognostic models based on The Cancer Genome Atlas Thyroid Cancer (TCGA-THCA) data, and to evaluate the association between tumor immune microenvironment and the prognostic model.MethodThe differentially expressed genes (DEGs) and tumor evolution were analyzed by scRNA-seq based on public databases. The potential regulatory networks of DEGs related to prognosis were analyzed by multi-omics data in the THCA. Logistic regression and Cox proportional hazards regression were utilized to construct the diagnosis and prognostic model of PTC. The performance of the diagnostic model was verified by bulk RNA sequencing (RNA-seq) of our cohort. The tumor immune microenvironment associated with the prognostic model was evaluated using multi-omics data. In addition, qRT-PCR was performed on tumor tissues and adjacent normal tissues of 20 patients to verify the expression levels of DEGs.ResultsThe DEGs screened by scRNA-seq can distinguish between tumor and healthy samples. DEGs play different roles in the evolution from normal epithelial cells to malignant cells. Three DEGs ((FN1, CLU, and ANXA1)) related to prognosis were filtered, which may be regulated by DNA methylation, RNA methylation (m6A) and upstream transcription factors. The area under curve (AUC) of the diagnostic model based on 3-gene in the validation of our RNA-seq was 1. In the prognostic model based on 3-gene, the overall survival (OS) of high-risk patients was shorter. Combined with the clinical information of patients, a nomogram was constructed by using tumor size (pT) and risk score to quantify the prognostic risk. The age and tumor size of high-risk patients in the prognostic model were greater. In addition, the increase of tumor mutation burden (TMB) and diversity of T cell receptor (TCR), and the decrease of CD8+ T cells in high-risk group suggest the existence of immunosuppressive microenvironment.ConclusionWe applied the scRNA-seq pipeline to focus on epithelial cells in PTC, simulated the process of tumor evolution, and revealed a prognostic prediction model based on 3 genes, which is related to tumor immune microenvironment.

Objective: To study the characteristics of the T cell receptor (TCR) repertoire in cancer tissue, peripheral blood and regional lymph nodes (LNs) from patients with papillary thyroid carcinoma (PTC).Methods: PTC tissue, peripheral blood mononuclear cells (PBMCs) and regional LNs of six patients with papillary thyroid carcinoma were harvested. T cell receptor beta-chain (TCRβ) profiling was performed though high-throughput sequencing (HTS), and IMonitor, MiXCR and VDJtools were used to analyze the characteristics of the TCR repertoire.Results: The results of IMonitor and those of MiXCR and VDJtools were very similar. The unique CDR3 of TCRβ from LNs was higher than that of PBMCs, and the CDR3 of TCRβ from LNs was higher than that of PTC tissue. Shannon's diversity index, D50, inverse Simpson index_mean and normalized Shannon's diversity index_mean of CDR3 from LNs were higher than those of PTCs and PBMCs. The HEC (high expansion clones) rate of CDR3 sequences at the amino acid level in PTC tissue was higher than that of PBMCs, which was higher than that of LNs. The V-J HEC rate of CDR3 was highest in PTC tissue, followed by PBMCs and LNs.Conclusion: TCR CDR3 profiling showed differences among and within the PBMCs, PTC tissues and regional LNs of PTC, including unique CDR3, CDR3 HEC at the amino acid level, CDR3 V-J HEC at the amino acid level, Shannon's diversity index and D50. The TCRβ repertoire of PTC tissue, peripheral blood and regional LNs of PTC provide a reference for further study of immunity mechanisms against PTC.

Colorectal cancer (CRC) with BRAF (V600E) is associated with microsatellite instability (MSI) that predicts response to immune checkpoint inhibitors. We demonstrated the interrogation of TCGA RNA-seq human datasets revealed that BRAFV600E tumors had significantly higher Programmed Death Ligand 1 (PD-L1) mRNA compared to non-mutated BRAF CRCs. Also, MSI-H tumors were evaluated as higher PD-L1 than MSS CRCs. Inhibition of MEK/ERK by cobimetinib or CDK inhibitor dinaciclib was shown to attenuate mutant BRAF-induced PD-L1 coincident with reduced c-JUN and YAP expression whose combined knockdown reduced PD-L1. Using TCGA datasets, PD-L1 mRNA expression in human colon cancers was significantly associated with YAP expression. The deletion of PD-L1 can reduce tumor cell growth shown by clonogenic assay. Analysis of the role of PD-L1 as a mediator of chemosensitivity was then performed. Knockout of PD-L1 was shown to attenuate the induction of DNA double-strand breaks (pH2AX) and caspase-3 cleavage by 5fluorouracil (5-FU) and paclitaxel compared to parental CRC cells. Results were confirmed in PD-L1 knockout MC38 murine CRC cells where re-expression of wild-type PD-L1 promoted DNA damage and apoptosis. We also performed the clonogenic assay and flow cytometry to prove that loss of PD-L1 attenuated DNA damage and apoptosis induced by diverse anti-cancer drugs that could be reversed by restoration of wild-type PD-L1. Mechanistically, knockout of PD-L1 reduced chemosensitivity in association with reductions in p-AKT and in BH3-only proteins BIM and BIK, rather than STAT3 in CRC cells. However, STAT3 had a significant role in melanoma, which shows the heterogeneity of cancers. In summary, BRAFV600E can upregulate PD-L1 expression that was induced by c-jun and YAP to enhance chemotherapy-induced apoptosis. Together, we demonstrate a potential role for PD-L1 as a regulator of chemotherapy-induced apoptosis whose deletion or suppression confers chemoresistance. These findings expand the understanding of PD-L1 functions to include nonimmune mechanisms and suggest the potential use of PD-L1 as a biomarker of response to cytotoxic chemotherapy.