In the area of large-scale graph data representation and semi-supervised learning, deep graph-based convolutional neural networks have been widely applied. However, typical graph convolutional network (GCN) aggregates information of neighbor nodes based on binary neighborhood similarity (adjacency matrix). It treats all neighbor nodes of one node equally, which does not suppress the influence of dissimilar neighbor nodes. In this paper, we investigate GCN based on similarity matrix instead of adjacency matrix of graph nodes. Gaussian heat kernel similarity in Euclidean space is first adopted, which is named EGCN. Then biologically inspired manifold similarity is trained in reproducing kernel Hilbert space (RKHS), based on which a manifold GCN (named MGCN) is proposed for graph data representation and semi-supervised learning with four different kernel types. The proposed method is evaluated with extensive experiments on four benchmark document citation network datasets. The objective function of manifold similarity learning converges very quickly on different datasets using various kernel functions. Compared with state-of-the-art methods, our method is very competitive in terms of graph node recognition accuracy. In particular, the recognition rates of MGCN (Gaussian kernel) and MGCN (Polynomial Kernel) outperform that of typical GCN about 3.8% on Cora dataset, 3.5% on Citeseer dataset, 1.3% on Pubmed dataset and 4% on Cora_ML dataset, respectively. Although the proposed MGCN is relatively simple and easy to implement, it can discover local manifold structure by manifold similarity learning and suppress the influence of dissimilar neighbor nodes, which shows the effectiveness of the proposed MGCN.
BackgroundGemcitabine can alter the immunogenic microenvironments, and the effect of gemcitabine plus programmed death-1 (PD-1)/ programmed death-ligand 1 (PD-L1) blockade in hepatocellular carcinoma (HCC) is investigated.MethodsSubcutaneous H22-green fluorescent protein (GFP) cells inoculation model was constructed and treated with gemcitabine, anti-PD-1 antibody (αPD-1), or the combination every four days when the tumor volumes reached about 50 mm3. Four days after the final treatment, primary tumor tissues were resected and dissociated, which were further subcutaneously injected on the contralateral side to construct the HCC relapse model. The infiltrated proportion of immune cells and PD-1 expression were quantified by flow cytometry. The relative content of transforming growth factor (TGF)-β, interleukin (IL)-12p70, and interferon (IFN)-γ were detected by the enzyme-linked immunosorbent assay (ELISA). Tumor volume and the number of tumor-free mice were evaluated.ResultsGemcitabine treatment can effectively increase the total proportion of infiltrating immune cells, reduce the proportion of myeloid-derived suppressor cells (MDSCs) and macrophages, and increase T cells proportion without significant growth inhibition. While after gemcitabine treatment, PD-L1 expression on tumor cells and PD-1 on T cells were significantly up-regulated. Subcutaneous tumors volume were reduced considerably after gemcitabine plus αPD-1 treatment compared with gemcitabine (P<0.01) or αPD-1 monotherapy (P<0.001) with the increased proportion of IL-2+CD8+T, CD8+T central memory cells (TCM), CD4 TCM, up-regulated IL12p70 and IFN-γ secretion, and down-regulated TGF-β. Gemcitabine plus αPD-1 blockade could inhibit the relapse tumor model as indicated with down-regulated tumor volume and increased number of tumor-free mice.ConclusionGemcitabine up-regulates the proportion of intratumor CD8+T and the relative expression of PD-1/PD-L1, and the combination of PD-1/PD-L1 blockade can further inhibit the growth and the relapse of HCC. 相似文献
Acute myeloid leukemia (AML) is the most common hematological malignancy in adults, but the etiology of it remains poorly understood. IL-35 is a recently described cytokine composed of an IL-12 subunit p35 and an IL-27 subunit Epstein–Barr virus induced gene 3 (EBI3), and has an immunosuppressive effect on inflammation through induction of regulatory T cells (Tregs) and suppression of Th1 and Th17. Recently, we have illustrated that concentrations of IL-35 in peripheral blood are up-regulated in newly diagnosed (ND) AML patients. However, whether IL-35 in bone marrow is increased in AML patients is not clear. In this study, we examined IL-35 in bone marrow by various methods including RT-PCR, ELISA, FCM and IHC, and found that IL-35 levels are also increased significantly in bone marrow of adult AML patients. Furthermore, we investigated that concentrations of bone marrow IL-35 in ND group were higher than that in complete remission (CR) group and control group, but there was no significant difference compared to that in relapse group. In conclusion, IL-35 was elevated in bone marrow of adult AML patients and this increase was correlated with the clinical stages of malignancy, suggesting that IL-35 is involved in pathogenesis of AML. 相似文献