Search

Background: Whether sex and co-mutations impact prognosis of patients with SMARCA4-mutated (mutSMARCA4) malignancies is not clear. Methods: This cohort included patients from Northern California Kaiser Permanente with next-generation sequencing (NGS) performed from August 2020 to October 2022. We used Cox regression modeling to examine the association between sex and overall survival (OS), adjusting for demographics, performance status, Charlson comorbidity index, receipt of treatment, tumor mutation burden (TMB), and TP53, KRAS, CDKN2A, STK11, and Keap1 co-mutations. Results: Out of 9221 cases with NGS performed, 125 cases (1.4%) had a mutSMARCA4. The most common malignancies with a mutSMARCA4 were non-small cell lung cancer (NSCLC, 35.2%), esophageal and stomach adenocarcinoma (12.8%), and cancer of unknown primary (11.2%). The most common co-mutations were p53 (mutp53, 59.2%), KRAS (mutKRAS, 28.8%), CDKN2A (mutCDKN2A, 31.2%), STK11 (mutSTK11, 12.8%), and Keap1 (mutKeap1, 8.8%) mutations. Male patients had substantially worse OS than female patients both among the entire mutSMARCA4 cohort (HR = 1.71, [95% CI 0.92–3.18]) with a median OS of 3.0 versus 43.3 months (p < 0.001), and among the NSCLC subgroup (HR = 14.2, [95% CI 2.76–73.4]) with a median OS of 2.75 months versus un-estimable (p = 0.02). Among all patients with mutSMARCA4, mutp53 versus wtp53 (HR = 2.12, [95% CI 1.04–4.29]) and mutSTK11 versus wtSTK11 (HR = 2.59, [95% CI 0.87–7.73]) were associated with worse OS. Among the NSCLC subgroup, mutp53 versus wtp53 (HR = 0.35, [0.06–1.97]) and mutKRAS versus wtKRAS (HR = 0.04, [0.003-.45]) were associated with better OS, while mutCDKN2A versus wtCDKN2A (HR = 5.04, [1.12–22.32]), mutSTK11 versus wtSTK11 (HR = 13.10, [95% CI 1.16–148.26]), and mutKeap1 versus wtKeap1 (HR = 5.06, [95% CI 0.89–26.61}) were associated with worse OS. Conclusion: In our cohort of patients with mutSMARCA4, males had substantially worse prognosis than females, while mutTP53, mutKRAS, mutCDKN2A, mutSTK11 and mutKeap1were differentially associated with prognosis among all patients and among the NSCLC subgroup. Our results, if confirmed, could suggest potentially unidentified mechanisms that underly this sex and co-mutation-dependent prognostic disparity among patients whose tumor bears a mutSMARCA4.

PURPOSE To examine the impact of TP53 gain-of-function (GOF) and non-GOF mutations on prognosis of advanced pancreatic ductal adenocarcinoma (PDAC) among patients with KRAS, CDKN2A, and SMAD4 comutations. METHODS This cohort included patients with locally advanced, recurrent, and de novo metastatic PDAC with next-generation sequencing performed from November 2017 to May 2022. We defined R175H, R248W, R248Q, R249S, R273H, R273L, and R282W as GOF and all other p53 mutations (mutp53) as non-GOF. We used Cox regression modeling to examine the association between GOF and non-GOF mutp53 and overall survival (OS), adjusting for demographics, performance status, Charlson comorbidity index, receipt of chemotherapy, and KRAS, CDKN2A, and SMAD4 comutations. RESULTS Of 893 total eligible patients, 68.5% had tumors with mutp53, 90.1% had KRAS mutations (mutKRAS), 44.7% had CDKN2A mutations (mutCDKN2A), and 17.0% had SMAD4 mutations. Among patients with mutp53, 121 had GOF and 491 had non-GOF. GOF mutp53 was associated with worse OS than non-GOF mutp53 (hazard ratio [HR], 1.27; 95% CI, 1.02 to 1.59) and wild-type p53 (wtp53; HR, 1.24; 95% CI, 0.98 to 1.57), whereas non-GOF was not associated with worse OS than wtp53 (HR, 0.95; 95% CI, 0.80 to 1.13). In addition, mutKRAS was associated with worse OS than wild-type KRAS in patients with mutCDKN2A (HR, 1.57; 95% CI, 0.88 to 2.80) but not in patients with wild-type CDKN2A (HR, 1.03; 95% CI, 0.76 to 1.39). CONCLUSION GOF and non-GOF mutp53 were associated with differential prognosis in advanced PDAC. The adverse effect of mutKRAS on OS appeared to be primarily driven by patients with mutCDKN2A. Our results provide new insight that could be helpful for prognostic stratification in clinical practice and for aiding future clinical trial designs.

Colorectal cancer (CRC) is one of the most common gastrointestinal malignancies. There are few recurrence risk signatures for CRC patients. Single-cell RNA-sequencing (scRNA-seq) provides a high resolution platform for prognostic signature detection. However, scRNA-seq is not practical in large cohorts due to its high cost and most single-cell experiments lack clinical phenotype information. Few studies have been reported to use external bulk transcriptome with survival time to guide the detection of key cell subtypes in scRNA-seq data. We proposed a data analysis framework to prioritize prognostic-associated subpopulations based on relative expression orderings (REOs). Cell type specific gene pairs (C-GPs) were identified to evaluate prognostic value for each cell type. We found REOs-based signatures could accurately classify most cell subtypes. C-GPs achieves higher precision compared with four current methods. Moreover, we developed single-cell gene pair signatures to predict recurrence risk for patients individually. Fibro_SGK1 cells and IgA+ IGLC2+ B cells were novel prognostic-associated subpopulations. A user-friendly toolkit, scRankXMBD(https://github.com/xmuyulab/scRank-XMBD), was developed to enable implementation of this framework. Our work facilitate the application of the rank-based method in scRNA-seq data for prognostic biomarker discovery and precision oncology.

Background & AimsLiver cancer is one of the most leading causes of cancer deaths. Cirrhosis is an important risk factor for liver cancer, which is the result of over-fibrosis caused by diffuse and long-term liver damage. Despite extensive research, a systematic study for characterizing similarity between liver cancer and cirrhosis at single cell resolution is still lacking.MethodsWe established a data analysis framework to elucidate cell lineage similarity between liver cancer and cirrhosis to discover prognostic-associated subpopulations. We integrated single-cell transcriptome data from liver samples at normal, cirrhotic and cancer conditions, which totally contained 78,000 cells. Gene regulation analysis, cellular interactions and trajectory analysis were performed to characterize cirrhosis-like cell subpopulations. Bulk transcriptomes were used to discover prognostic-associated subpopulations.ResultsBy aligning cellular subpopulations across different samples, we found remarkable similarity betweenKNG1+hepatocytes in cirrhosis andPGA5+hepatocytes in HCC. Furthermore, gene regulation analysis and cellular interactions implicated E2F1, FOXA2, EGF, CDH and ANGPTL signaling in maintaining cirrhosis-like ecosystem. Strikingly, subpopulations with higher expression of cirrhosis-like signatures were associated with patients’ worse survival.ConclusionsWe revealed a previously unexplored cirrhosis-like ecosystem of liver cancer, which could provide novel biomarkers for therapeutic interventions in HCC. Core analysis modules in this study were integrated into a user-friendly toolkit, SIMscRNA(https://github.com/xmuhuanglab/SIM-scRNA), which could facilitate the exploration of similarity and heterogeneity between precancerous diseases and solid tumors.

Background Targeted therapy for non-small cell lung cancer is histology dependent. However, histological classification by routine pathological assessment with hematoxylin-eosin staining and immunostaining for poorly differentiated tumors, particularly those from small biopsies, is still challenging. Additionally, the effectiveness of immunomarkers is limited by technical inconsistencies of immunostaining and lack of standardization for staining interpretation. Results Using gene expression profiles of pathologically-determined lung adenocarcinomas and squamous cell carcinomas, denoted as pADC and pSCC respectively, we developed a qualitative transcriptional signature, based on the within-sample relative gene expression orderings (REOs) of gene pairs, to distinguish ADC from SCC. The signature consists of two genes, KRT5 and AGR2, which has the stable REO pattern of KRT5 > AGR2 in pSCC and KRT5 AGR2 in pADC. In the two test datasets with relative unambiguous NSCLC types, the apparent accuracy of the signature were 94.44 and 98.41%, respectively. In the other integrated dataset for frozen tissues, the signature reclassified 4.22% of the 805 pADC patients as SCC and 12% of the 125 pSCC patients as ADC. Similar results were observed in the clinical challenging cases, including FFPE specimens, mixed tumors, small biopsy specimens and poorly differentiated specimens. The survival analyses showed that the pADC patients reclassified as SCC had significantly shorter overall survival than the signature-confirmed pADC patients (log-rank p = 0.0123, HR = 1.89), consisting with the knowledge that SCC patients suffer poor prognoses than ADC patients. The proliferative activity, subtype-specific marker genes and consensus clustering analyses also supported the correctness of our signature. Conclusions The non-subjective qualitative REOs signature could effectively distinguish ADC from SCC, which would be an auxiliary test for the pathological assessment of the ambiguous cases.

Dense-array concentrating photovoltaic (DA-CPV) systems suffer from power generation limits due to extreme operation conditions. This study primarily aims to analyze the multi-physics effects of various optical, electrical, and thermal conditions on the performance of DA-CPV systems. Different module configurations are employed in the simulations to evaluate the coupling impacts of changing illumination distributions and temperature profiles on the system performance. The results demonstrate that module configurations have significant influence on the output power and temperature contour of DA-CPV modules. Compared with conventional total-cross-tied connections, decrements of average module temperature and central module temperature by at least 5 and 12 °C, respectively, and increment of output power by at least 48.29% are achieved based on the quartered rotational symmetry (QRS) connection. Under various non-uniform illuminations, the maximum power will be obtained at anti-Gaussian temperature profiles. Compared with the uniform and Gaussian temperature profiles, the inverse Gaussian profiles can further improve the module output power by at most 1%. Meanwhile, relying on the achieved correlation between the optimized temperature profiles and illumination shapes for the CPV module with the QRS connection, the output performance of the CPV module under various raidation intensities is evaluated. In the range of 100–1200 W/m2 of solar radiation, more than 99.7% of the maximum module power is maintained with the optimized anti-Gaussian temperature profile.

e16294 Background: Developmentally pancreas head derives from ventral bud while pancreas neck, body and tail derive from dorsal bud. Pancreatic ductal adenocarcinoma (PDAC) frequently harbors multiple mutations including KRAS, TP53, CDKN2A, and others. It is unknown how TP53 gain-of-function (GOF) and non-gain-of-function (non-GOF) mutations affect the prognosis. Methods: We retrospectively examined a cohort of 741 Kaiser Permanente (KP) patients with locally advanced/metastatic PDAC who had NGS performed to determine the association of KRAS (mutKRAS), TP53 (mutp53) and CDKN2A (mutCDKN2A) mutations (individually and in combination) with overall survival (OS). We used Cox modeling to estimate hazard ratios (HR) adjusted for age, sex, ethnicity, performance status, Charlston Comorbidity Index, chemotherapy received, anatomic location and co-mutations. We also analyzed the TCGA PDAC dataset to examine the association of OS with these same mutations. Results: In the KP cohort, patient ages ranged from 36 to 94 years and approximately 50% were female. In 384 patients PDAC was on the head, and 357 patients had PDAC on a non-head location (neck, body, and tail). Those with head PDAC had modestly better OS compared to non-head PDAC (HR = 0.87), and this appeared to be driven by the subsets of patients with wtp53 (HR = 0.68), with wtKRAS (HR = .74) and with wtCDKN2A (HR = .78). Approximately 67.5% of patients had mutp53, 89.2% had mutKRAS and 44.8% had mutCDKN2A. Among all KP patients, OS was similar for patients with mutp53 vs. wtp53 (HR = 1.05); worse for patients with mutKRAS vs. wtKRAS (HR = 1.26), and worse for patients with mutCDKN2 vs. wtCDKN2A (HR = 1.51). Interestingly, among patients with a GOF mutp53, those with mutCDKN2A had substantially worse OS vs patients with wtCDKN2A (HR = 2.56, 95% CI 1.46-4.50). In contrast, among patients with a non-GOF mutp53, patients with mutCDKN2A had only moderately worse OS compared to patients with wtCDKN2A (HR = 1.37, 95% CI 1.06-1.79). Analysis of the TCGA PDAC dataset showed that the number of mutations (0, 1, 2, or 3, of p53, KRAS and CDKN2A) was associated with incrementally worse OS ( p < .001). Conclusions: Better OS of head vs. non-head PDAC was primarily driven by patients with wtp53, wtKRAS, and wtCDKN2A. The adverse effect of mutCDKN2A on OS appears to be most pronounced in patients with GOF mutp53. Our TCGA analysis suggests interactions among TP53, KRAS and CDKN2A mutations in affecting PDAC survival.

RNA-sequencing enables accurate and low-cost transcriptome-wide detection. However, expression estimates vary as reference genomes and gene annotations are updated, confounding existing expression-based prognostic signatures. Herein, prognostic 9-gene pair signature (GPS) was applied to 197 patients with stage I lung adenocarcinoma derived from previous and latest data from The Cancer Genome Atlas (TCGA) processed with different reference genomes and annotations. For 9-GPS, 6.6% of patients exhibited discordant risk classifications between the two TCGA versions. Similar results were observed for other prognostic signatures, including IRGPI, 15-gene and ORACLE. We found that conflicting annotations for gene length and overlap were the major cause of their discordant risk classification. Therefore, we constructed a prognostic 40-GPS based on stable genes across GENCODE v20-v30 and validated it using public data of 471 stage I samples (log-rank P

Additional file 3: Table S2. Technical details of TMB evaluation.

Background Human pancreatic ductal adenocarcinoma (PDAC) responds poorly to immune checkpoint inhibitor (ICPi). While the mechanism is not completely clear, it has been recognized that tumor microenvironment (TME) plays key roles. We investigated if targeting CD47 with a monoclonal antibody could enhance the response of PDAC to ICPi by altering the TME. Methods Using immunohistochemistry, we examined tumor-infiltrating CD68+ pan-macrophages (CD68+ M) and CD163+ M2 macrophages (CD163+ M2) and tumor expression of CD47 and PD-L1 proteins in 106 cases of PDAC. The efficacy of CD47 blockade was examined in xenograft models. CD45+ immune cells from syngeneic tumor models were subjected to single-cell RNA-sequencing (scRNA-seq) by using the 10x Genomics pipeline. Results We found that CD47 expression correlated with the level of CD68+ M but not CD163+ M2. High levels of tumor-infiltrating CD68+ M, CD163+ M2, and CD47 expression were significantly associated with worse survival. CD47high/CD68+ Mhigh and CD47high/CD163+ M2high correlated significantly with shorter survival, whereas CD47low/CD68+ Mlow and CD47low/CD163+ M2low correlated with longer survival. Intriguingly, CD47 blockade decreased the tumor burden in the Panc02 but not in the MPC-83 syngeneic mouse model. Using scRNA-seq, we showed that anti-CD47 treatment significantly remodeled the intratumoral lymphocyte and macrophage compartments in Panc02 tumor-bearing mice by increasing the pro-inflammatory macrophages that exhibit anti-tumor function, while reducing the anti-inflammatory macrophages. Moreover, CD47 blockade not only increased the number of intratumoral CD8+ T cells, but also remodeled the T cell cluster toward a more activated one. Further, combination therapy targeting both CD47 and PD-L1 resulted in synergistic inhibition of PDAC growth in the MPC-83 but not in Panc02 model. MPC-83 but not Panc02 mice treated with both anti-CD47 and anti-PD-L1 showed increased number of PD-1+CD8+ T cells and enhanced expression of key immune activating genes. Conclusion Our data indicate that CD47 targeting induces compartmental remodeling of tumor-infiltrating immune cells of the TME in PDAC. Different PDAC mouse models exhibited differential response to the anti-CD47 and anti-PD-L1 blockade due to the differential effect of this combination treatment on the infiltrating immune cells and key immune activating genes in the TME established by the different PDAC cell lines.

Background Immune checkpoint inhibitors are effective in some cases of lung adenocarcinoma (LUAD). Whole-exome sequencing has revealed that the tumour mutation burden (TMB) is associated with clinical benefits among patients from immune checkpoint inhibitors. Several commercial mutation panels have been developed for estimating the TMB regardless of the cancer type. However, different cancer types have different mutational landscapes; hence, this study aimed to develop a small cancer-type-specific mutation panel for high-accuracy estimation of the TMB of LUAD patients. Methods We developed a small cancer-type-specific mutation panel based on coding sequences (CDSs) rather than genes, for LUAD patients. Using somatic CDSs mutation data from 486 LUAD patients in The Cancer Genome Atlas (TCGA) database, we pre-selected a set of CDSs with mutation states significantly correlated with the TMB, from which we selected a CDS mutation panel with a panel-score most significantly correlated with the TMB, using a genetic algorithm. Results A mutation panel containing 106 CDSs of 100 genes with only 0.34 Mb was developed, whose length was much shorter than current commercial mutation panels of 0.80–0.92 Mb. The correlation of this panel with the TMB was validated in two independent LUAD datasets with progression-free survival data for patients treated with nivolumab plus ipilimumab and pembrolizumab immunotherapies, respectively. In both test datasets, survival analyses revealed that patients with a high TMB predicted via the 106-CDS mutation panel with a cut-point of 6.20 mutations per megabase, median panel score in the training dataset, had a significantly longer progression-free survival than those with a low predicted TMB (log-rank p = 0.0018, HR = 3.35, 95% CI 1.51–7.42; log-rank p = 0.0020, HR = 5.06, 95% CI 1.63–15.69). This small panel better predicted the efficacy of immunotherapy than current commercial mutation panels. Conclusions The small-CDS mutation panel of only 0.34 Mb is superior to current commercial mutation panels and can better predict the efficacy of immunotherapy for LUAD patients, and its low cost and time-intensiveness make it more suitable for clinical applications.

Additional file 1: Table S1. Clinical and pathologic features of PDAC patients. Table S2. Univariate and multivariate Cox proportional analysis for overall survival. Table S3. Multivariate analysis of CD47 expression combined with PD-L1 expression or TAM in PDAC patients. Analysis was adjusted for diameter, grade, and TNM stage. Figure S1. Tumor-infiltrating CD45+ immune cells in Panc02 tumor-bearing mice for single-cell analysis. Figure S2. Immunostaining of CD4, CD8, iNOS, and CD206 in Panc02 tumors. Figure S3. Heatmap from scRNA-seq displaying expression of select genes in each lymphoid cell subpopulation.

Transformer model based on self-attention mechanism [17] has achieved state-of-the-art in recent evaluations. However, it is still unclear how much room there is for improvement of the translation system based on this model. In this paper we further explore how to build a stronger neural machine system from four aspects, including architectural improvements, diverse ensemble decoding, reranking, and post-processing. Experimental results on CWMT-18 Chinese \(\leftrightarrow \) English tasks show that our approach can consistently improve the translation performance of 2.3–3.8 BLEU points than the strong baseline. Particularly, we find that ensemble decoding with a large number of diverse models is crucial for significant improvement.

Additional file 1: Table S1. Clinical characteristics of patients treated with curative surgery resection only. Table S2. The top 50 subtype-opposite gene pairs. Table S3. The histological classification of samples classified by pathological assessment, signature and clustering in the GSE50081. Table S4. The histological classification of samples classified by pathological assessment, signature and cluster in the GSE58661. Figure S1. Kaplan-Meier curves of OS respectively for the SCC and ADC groups reclassified by the signature and original pathological assessment in the data integrated by 6 datasets. Figure S2. The consensus clustering of the samples based on top 1000 most variable genes in the GSE50081 dataset. Figure S3. The consensus clustering of the samples based on top 2000 most variable genes in the GSE50081 dataset. Figure S4. The consensus clustering of the samples based on top 3000 most variable genes in the GSE50081 dataset. Figure S5. The consensus clustering of the samples based on top 1000 most variable genes in the GSE58661 dataset for small biopsy specimens. Figure S6. The consensus clustering of the samples based on top 2000 most variable genes in the GSE58661 dataset for small biopsy specimens. Figure S7. The consensus clustering of the samples based on top 3000 most variable genes in the GSE58661 dataset for small biopsy specimens. Figure S8. The Kaplan-Meier curves of overall survival respectively for the ADC and SCC groups of patients treated with curative surgery resection only.

Nanofluids (NFs) have been investigated as optical filters for photovoltaic/thermal applications, since they are more effective to produce heat with higher temperatures. However, the effect of temperature on both the stability and optical property of NFs has been little reported. In this work, we synthesized the SiO2-H2O NFs under different conditions and different nanoparticle volume fractions, and investigated the NF dispersion stability and optical properties at elevated temperatures. We analyzed the effects of several factors including the sonication temperature, the exposure temperature, the nanoparticle volume fraction and the exposure time. The results showed that reducing the maximum sonicating temperature during the NF fabrication process is beneficial to prolonging the NF stability, and the prepared NF stability was rather good with the maximum sonicating temperature of 30 °C. The absorptance of NFs decreased when the exposure temperature increased from 25 to 90 °C, and agglomeration more likely occurred for higher exposure temperatures. It is also found that the adding of nanoparticles helped to increase the spectrum absorption ability of NF. As a result, the overall energy efficiencies of the photovoltaic/thermal device using the NFs as the optical filter can be increased compared with that using the water filter.

Concentrated photovoltaic/thermal hybrid collectors have received ever-increasing attention due to the simultaneous output of electric and thermal energy. For further improvement of concentrated hybrid PV/T systems, a simulation method combining the multiphysics fields is necessary to accurately analyze the optical, thermal, and electric performance. Herein, a three-dimensional numerical study has been conducted on a low concentrated photovoltaic/thermal system utilizing a heat transfer fluid as the cooling medium and a compound parabolic concentrator as the mirror field. A finite volume (FV)-CFD code has been employed to simulate the entire model, where the optical modelling is validated theoretically with the Monte Carlo ray-tracing method. The influences of employing various heatsink designs (U-type and Z-type) and coolants (water, ethylene glycol, and therminol VP-1) are numerically investigated. The economic feasibility of the hybrid PV/T system is also assessed in comparison with the standalone PV-cell. Good compatibility with the empirical data was obtained when the appropriate modelling tunings were applied. It is also shown that, on a typical day, the total energy and exergy efficiencies of the system are up to 57.66% and 7.94%, respectively. The Z-type heatsink decreases the average PV-cell temperature than the U-type design, and also the output power is slightly enhanced.

Concentrated photovoltaic/thermal (CPVT) hybrid systems can provide a simultaneous supply of electrical and thermal energy. The performance of CPVT system based on liquid absorption spectral beam filter with different structure configurations is theoretically investigated in this paper. Main efforts are put to reveal the effects of vacuum enclosure surrounding the optical filter and the heat transfer medium between the thermal receiver and the solar cells on the system performance at different operating conditions. The effect of long-term operation under high-temperature exposure is also analyzed. The simulation results show that both the vacuum enclosure around the thermal unit and the heat transfer medium between the thermal receiver and solar cells can influence the transmitted energy arriving the solar cells and the temperature of the whole module, which are very sensitive to the electrical and thermal performance. The use of vacuum enclosure decreases the electrical efficiency by about 4%, while it leads to a much higher thermal efficiency. With large coolant flow rates, the configuration without the vacuum enclosure shows both higher overall energy and exergetic efficiencies. It is also suggested to use an air gap between the receiver and PV cells instead of a vacuum layer or a solid layer.

This paper describes the submission of the NiuTrans neural machine translation system for the WMT 2018 Chinese ↔ English news translation tasks. Our baseline systems are based on the Transformer architecture. We further improve the translation performance 2.4-2.6 BLEU points from four aspects, including architectural improvements, diverse ensemble decoding, reranking, and post-processing. Among constrained submissions, we rank 2nd out of 16 submitted systems on Chinese → English task and 3rd out of 16 on English → Chinese task, respectively.

Table S1. The primer sequences used for polymerase chain reaction. Table S2. The nucleotide sequence of the primers used for qRT-PCR. Table S3. The sequence of miR-200c mimic, miR-141 mimic, anti-miR-200c mimic (Has-miR-200c inhibitor), and anti-miR-141 mimic (Has-miR-141 inhibitor) used for lentivirus transfection and luciferase reporter assay. Table S4. Characteristics of patients with pancreatic cancer (N = 37). Table S5. Characteristics of patients with pancreatic cancer from the TCGA database (N = 177). Figure S1. Identifying miR-141/200c target genes using the TargetScan software program. Figure S2. The levels of CpG methylation of the promoter region of miR-200a/200b/429 in PDAC. Figure S3. Lentiviral expression of miR-141 and miR-200c and their inhibitors in pancreatic cancer cell lines. Figure S4. The effect of miR-141 and miR-200c inhibitors on cell migration and invasion in vitro and tumor growth in xenograft. Figure S5. miR-141 and miR-200c inhibit the expression of WIPF1 in HPDE cell line. Figure S6. Lentiviral expression of shWIPF1 in pancreatic cancer cell lines. Figure S7. WIPF1 antagonizes the inhibitory effect of miR-141/200c on cell migration, invasion and metastasis of PDAC. (DOCX 17513 kb)

In the multi-label classification task (Automatic Tagging of Zhihu Questions), we present a classification system which includes five processes. Firstly, we use a preprocessing step to solve the problem that there is too much noise in the training dataset. Secondly, we choose several neural network models which proved effective in text classification task. Then we introduce k-max pooling structure to these models to fit this task. Thirdly, in order to obtain a better performance in ensemble process, we use an experiment-designing process to obtain classification results that are not similar to each other and all achieve relatively high scores. Fourthly, we use an ensemble process. Finally, we propose a method to estimate how many labels should be chosen. With these processes, our F1 score achieves 0.5194, which ranked No. 3.

Human pancreatic ductal adenocarcinoma (PDAC) exhibits marginal responses to anti-PD-1/PD-L1 immunotherapy and its mechanism remains poorly understood. We have investigated the effect of anti-PD-L1 and c-Myc inhibition in PDAC. Using 87 patients with PDAC from our hospital database we found a significant correlation between the expression of PD-L1 and c-Myc. Moreover, the expression of both PD-L1 and c-Myc was associated with poor overall survival. In addition, we confirmed this finding with the PDAC patients in the TCGA database. Using several PDAC cell lines we demonstrated a significant correlation between the expression of PD-L1 and c-Myc. We also found that expression of PD-L1 correlated with high-grade histology. JQ1, an inhibitor of c-Myc inhibited PD-L1 expression and tumor growth. Using xenograft models, we demonstrated that the combination of JQ1 and anti-PD-L1 antibody exerted synergistic inhibition of PDAC growth. Our data demonstrated that the expression of PD-L1 and c-Myc may be helpful prognostic biomarkers, and their inhibition may potentially serve as an effective treatment for PDAC.