Search

Background. Glioma is the most common primary intracranial tumor and is associated with poor prognosis. Identifying effective biomarkers for glioma is particularly important. MXRA5, a secreted glycoprotein, is involved in cell adhesion and extracellular matrix remodeling and has been reported to be expressed in many cancers. However, the role and mechanism of action of MXRA5 in gliomas remain unclear. This study was aimed at investigating the role of MXRA5 at the transcriptome level and its clinical prognostic value. Methods. In this study, RNA microarray data of 301 glioma patients from the Chinese Glioma Genome Atlas (CGGA) were collected as a training cohort and RNA-seq data of 702 glioma samples from The Cancer Genome Atlas (TCGA) were used for validation. We analyzed the clinical and molecular characteristics as well as the prognostic value of MXRA5 in glioma. In addition, the expression level of MXRA was evaluated in 28 glioma tissue samples. Results. We found that MXRA5 expression was significantly upregulated in high-grade gliomas and IDH wild-type gliomas compared to controls. Receiver operating characteristic (ROC) analysis showed that MXRA5 is a potential marker of the mesenchymal subtype of glioblastoma multiforme (GBM). We found that MXRA5 expression is highly correlated with immune checkpoint molecule expression levels and tumor-associated macrophage infiltration. High MXRA5 expression could be used as an independent indicator of poor prognosis in glioma patients. Conclusion. Our study suggests that MXRA5 expression is associated with the clinicopathologic features and poor prognosis of gliomas. MXRA5 may play an important role in the immunosuppressive microenvironment of glioma. As a secreted glycoprotein, MXRA5 is a potential circulating biomarker for glioma, deserving further investigation.

Inhibitor of nuclear factor kappa‐B kinase subunit epsilon (IKBKE) plays critical roles in the proliferation, invasion, and epithelial‐mesenchymal transition (EMT) of glioblastoma (GBM). However, as an immune response factor, few studies have focused on the role of IKBKE in the glioblastoma‐mediated immunosuppressive microenvironment. Here, we found a higher IKBKE expression level in gliomas corresponding to higher malignancy of the tumor. The highest level of IKBKE expression was examined in the core region of GBM tissues as well as the mesenchymal subtype, which are featured with necrosis, immunocyte infiltration, and immunosuppression. Further in silico analysis demonstrated that the JAK/STAT as the signaling pathway most associated with IKBKE in mesenchymal GBM. The co‐expression patterns of IKBKE, pSTAT3, and PD‐L1 were detected within GBM tissues. Mechanistically, IKBKE could interact with STAT3 and thus enhancing the phosphorylation level of STAT3 and its nuclear translocation. In addition, pSTAT3 could transcriptionally regulate the expression of PD‐L1 through binding to its promoter. In vivo results further confirmed the inhibitory effect of the IKBKE downregulation on tumor growth. Collectively, our findings suggest IKBKE as the central node in the crosstalk between NF‐κB and STAT3 signaling within mesenchymal GBM. Targeting GBM through inhibiting IKBKE could restrain tumor growth and tumor‐mediated immunosuppressive environment., 1. Higher IKBKE expression was detected in higher grades of glioma. 2. IKBKE is enriched in mesenchymal GBM, which confers immunosuppressive signature. 3. IKBKE correlates with JAK/STAT signaling in MES GBM. 4. IKBKE activates the phosphorylation of STAT3, which transcriptionally regulates PD‐L1 expression.

5038 Background: Accurate analysis of the genomic alteration landscape within tissue- and plasma-derived tumor DNA using next-generation sequencing (NGS) may provide insights into specific patient populations that benefit from different therapies. The interchangeable use of tissue- and plasma-based assessments may benefit patients when tissue availability is limited, a common occurrence in individuals with mCRPC. To understand the potential sources of technical and biological variability in this setting, we performed comprehensive comparative analyses across 3 NGS platforms, using samples from patients enrolled in CheckMate 9KD, a phase 2 study of nivolumab combined with docetaxel, rucaparib, or enzalutamide for patients with confirmed mCRPC (NCT03338790). Methods: We performed retrospective integrated analyses of sequence and structural alterations identified through comprehensive genomic profiling (CGP) of DNA obtained from formalin-fixed, paraffin-embedded tissue specimens and cell-free DNA obtained from plasma. Tissue-based analysis was performed using the FoundationOne assay (F1, 395 genes), while the FoundationACT (FACT, 70 genes) and GuardantOMNI (OMNI, 500 genes) assays were used for plasma-based analysis. Analysis was performed on samples from 103 patients for which datasets from all 3 assays were available. Inter-platform analysis considered variants with ≥ 0.50% variant allele fraction and common to the shared pairwise targeted regions, while excluding synonymous variants. Results: Through broad profiling of DNA obtained from tissue and plasma, we uncovered previously identified recurrent alteration of AR, TP53, PTEN, and TMPRSS2 fusion with ETS genes. Additionally, we found that 42% (F1), 45% (FACT), and 34% (OMNI) of patients harbored a combination of germline and somatic mutations in homologous recombination repair pathway genes. Across all samples, median tumor mutational burden was 3.5 mutations per megabase (mut/Mb) by F1 and 8.6 mut/Mb by OMNI. Inter-platform variant analyses demonstrated concordance of 52% for F1 vs FACT, 40% for F1 vs OMNI, and 75% for FACT vs OMNI. Conclusions: Overall, these data demonstrate the value of integrated tissue and liquid biopsy profiling in mCRPC. Both technical and biological sources of variation, including panel size, mutation detection algorithms, variant annotation and reporting, analytical performance, circulating tumor DNA levels, and tumor heterogeneity, may be captured differently by tissue- and plasma-based techniques, accounting for the discordance in reported results. Clinical trial information: NCT03338790.

5044 Background: CheckMate 9KD is a phase 2 trial of NIVO (anti-PD-1) combined with either rucaparib, docetaxel, or enzalutamide for mCRPC. PARP inhibitors, like rucaparib, increase cellular DNA damage, particularly in tumors with DNA repair defects, leading to genomic instability and cell death. This DNA damage promotes immune priming and adaptive PD-L1 upregulation. Consequently, dual PD-(L)1 and PARP inhibition is a plausible therapeutic strategy for mCRPC. We report final results for cohort A1 (NIVO + rucaparib for post-CT mCRPC) of CheckMate 9KD. Methods: Cohort A1 enrolled patients (pts) with post-CT mCRPC (1–2 prior taxane regimens), ongoing ADT, ≤ 2 prior novel hormonal therapies (abiraterone, enzalutamide, etc) for mCRPC, and no prior PARP inhibitor treatment. Pts received NIVO 480 mg Q4W + rucaparib 600 mg BID until disease progression/unacceptable toxicity (NIVO dosing limited to 2 yrs). Coprimary endpoints: objective response rate (ORR) per PCWG3 criteria and prostate-specific antigen response rate (PSA-RR; ≥ 50% PSA reduction) in all treated pts and pts with homologous recombination deficiency positive (HRD+) tumors, determined before enrollment. Secondary endpoints included radiographic progression-free survival (rPFS), overall survival (OS), and safety. Results: Of 88 treated pts, median age 66 yrs (range, 46–85), 34.1% had visceral metastases and 65.9% had measurable disease. Median follow-up was 11.9 mo. Pts had a median of 4.5 NIVO doses and 3.8 mo of rucaparib. The table summarizes primary and key secondary efficacy results, and shows better outcomes for HRD+ vs HRD–/not evaluable (NE) tumors. In pts with BRCA2 mutations, confirmed ORR was 37.5% (3/8 pts) and confirmed PSA-RR was 45.5% (5/11 pts). Any-grade treatment-related AEs (TRAEs) occurred in 93.2% of pts, most commonly nausea (40.9%) and fatigue (33.0%). Grade ≥ 3 TRAEs occurred in 54.5% of pts, most commonly anemia (20.5%) and neutropenia (10.2%). TRAEs led to discontinuation in 27.3% of pts. One pt had a stroke, considered related to rucaparib by the investigator, after 28 days on rucaparib and 2 NIVO doses and died 2 months later due to post-thrombolysis hematoma. Conclusions: NIVO + rucaparib is active in pts with HRD+ post-CT mCRPC, although the trial design and short follow-up limit assessment of benefits of the combination vs individual components. Pts with HRD– tumors did not appear to benefit from either drug. No new safety signals were observed with NIVO + rucaparib. Additional biomarker analyses are ongoing. Clinical trial information: NCT03338790. [Table: see text]

Background: Breast cancer has been extensively characterized molecularly with cohorts such as The Cancer Genome Atlas (TCGA). However, as the standard of care evolves, there is a need for contemporaneous datasets with more detailed treatment and outcomes data. The ORIEN Avatar database (M2Gen, Tampa, FL) consists of clinical and tumor sequencing data from treatment-naive or on-treatment biopsies. This study evaluates the quality of ORIEN Avatar's RNA-sequencing (RNASeq) data, interrogates the impact of prior treatment on PAM50 subtyping, and reports associations between breast cancer subtypes and estimated immune cell populations. Methods: In the ORIEN Avatar dataset, RNASeq data were prepared using a multistep normalization process. We assessed the quality of normalized RNASeq data using 3 modalities. Using a principal-component (PC) analysis, we interrogated the association of gene expression with potential technical and biological factors, including processing batch effects, preservation methods, and treatment status. We evaluated concordance between immunohistochemistry (IHC)-based subtyping and intrinsic subtyping using the PAM50 gene expression signature. Prevalence of in silico-derived tumor-infiltrating immune cell compositions in ORIEN Avatar compared with TCGA breast cancer samples was investigated. Results: The ORIEN Avatar breast cancer cohort comprised 560 patients with RNASeq data. The first 3 PCs (accounting for 11%, 9%, and 7% of total explained variance, respectively) were associated with biological factors, including PAM50 subtyping and HER2 receptor and hormone receptor (HR) status. PAM50 subtype was largely concordant with IHC subtype, per clinical record of biomarker testing, and prior treatment status did not impact PAM50 subtyping. We observed that 68% of basal samples were triple negative by IHC, 92% of luminal samples were HER2-negative and HR-positive by IHC, and 62% of HER2-enriched samples per RNASeq were HER2 receptor positive by IHC. Several in silico-derived tumor-infiltrating immune cell compositions, including activated CD4 memory T cells, M0 macrophages, and activated dendritic cells, were enriched in triple-negative breast cancer samples, consistent with TCGA. Correlative analysis of RNASeq data with other biomarker data and clinical outcome is planned. Conclusions: Our results using real-world (RW) ORIEN Avatar RNASeq data were consistent with molecular profiling of breast cancer assessed using IHC and TCGA. This proof-of-concept study found no association between prior treatment status and PAM50 intrinsic subtyping. Our analysis provides a framework to assess the quality of RW molecular profiling and highlights the feasibility of leveraging harmonized molecular data to replicate and discover novel biological evidence. Citation Format: Xuya Wang, Bin Li, Peter M. Szabo, Han Chang, Mustimbo Roberts, Alice M. Walsh. Immune contexture of breast cancer subtypes in real-world molecular data [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2117.

Background: Anti–programmed death-1/programmed death ligand 1 inhibitors have been approved for the treatment of advanced small cell lung cancer (SCLC) as a single disease entity. However, efforts to detect and characterize biologically distinct subtypes of SCLC based on gene expression profiling (GEP) have identified multiple subtypes of SCLC, with potential therapeutic implications. These SCLC subtypes can be characterized as neuroendocrine (NE) or non-NE by GEP. Mutation in the retinoblastoma (Rb) transcriptional corepressor 1 (RB1) gene occurs in 65–85% of patients with SCLC, and RB1 mutation may be associated with improved survival with chemotherapy. In this retrospective, exploratory analysis, we investigated the association of genomic and transcriptional biomarkers, including RB1 mutation, with response to immuno-oncology (I-O) in patients with SCLC. Methods: Biomarker evaluation was performed in baseline samples from patients with SCLC who received second-line or later nivolumab ± ipilimumab (NIVO±IPI) in CheckMate 032 (NCT01928394). RB1 wild type (RB1WT) or mutant (RB1MUT) status and tumor mutational burden (TMB) were determined by whole exome sequencing (WES). Gene expression signatures for SCLC transcriptional subtype (6 genes), Rb protein loss of function (LOF; 87 genes), and inflammation (4 genes) were determined using RNA sequencing. Results: Of 460 total samples from patients with SCLC, WES and GEP data were available for 279 and 286 samples, respectively. 166 patients (59%) carried frameshift, in-frame insertions or deletions, truncations, or stop-gained mutations in RB1. No differences in RB1 mutation frequency were observed between treatment arms. Overall survival (OS) with NIVO was lower in RB1MUT patients compared with RB1WT (HR = 1.46 [1.02–2.1]). RB1 mutation status was not associated with OS in patients receiving NIVO+IPI. Similar trends were observed for objective response rates. Moreover, stratifying patients based on an Rb protein LOF gene expression signature showed an association between Rb protein loss and lower OS with NIVO (HR = 1.3 [1.05–1.65]) and NIVO+IPI (HR = 1.2 [0.95–1.49]). RB1 mutation status also associated with NE transcriptional subtype (P < 0.0001). Stratification of patients into NE/non-NE subtypes was not associated with OS. TMB and inflammation signatures did not differ between NE and non-NE SCLC subtype or RB1 mutation status. Conclusion: This analysis of SCLC samples suggests that transcriptionally subtyping SCLC has limited clinical utility. Rb protein function, inflammatory gene signatures, and TMB may represent independent biomarkers of response to I-O in patients with SCLC, warranting further investigation to determine the predictive value of these biomarkers, alone and in combination. Citation Format: David Balli, Xuya Wang, Naiyer A. Rizvi, Timothy A. Chan, Afshin Dowlati, Parul Doshi, Han Chang. Association of retinoblastoma function with response to immuno-oncology treatment in patients with small cell lung cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2047.

Background The antitumor activity of programmed death-1 (PD-1)/PD-1 ligand 1 (PD-L1) inhibition alone is limited in patients (pts) with metastatic castration-resistant prostate cancer (mCRPC). Combining immunotherapy with standard-of-care chemotherapy could improve outcomes in mCRPC. We report interim analysis results for the nivolumab + docetaxel (NIVO+DOCE) treatment arm from CheckMate 9KD. Methods CheckMate 9KD is a phase II study of NIVO in combination with DOCE, rucaparib, or enzalutamide. Eligible pts had confirmed metastatic adenocarcinoma of the prostate, ongoing androgen deprivation therapy, and evaluable tumor biopsy. Pts assigned to NIVO+DOCE (chemotherapy-naive and received ≤2 prior second-generation hormonal therapies) received NIVO 360mg every 3 weeks + DOCE 75mg/m2 every 3 weeks + prednisone 5mg twice daily for ≤10 cycles followed by NIVO 480mg every 4 weeks alone until disease progression or unacceptable toxicity (up to 2 years). Coprimary endpoints were objective response rate (ORR) and prostate-specific antigen response rate (PSA-RR; defined as≥50% PSA reduction from baseline). Secondary endpoints included radiographic progression-free survival (rPFS), and safety/tolerability in all treated pts. Association of biomarkers with efficacy was an exploratory endpoint. Results This interim analysis included 41 pts in the NIVO+DOCE arm with a minimum follow-up of 28 weeks, of whom 19 (46.3%) had measurable disease. At the time of the analysis, 24 (58.5%) had discontinued study treatment. ORR in pts with measurable disease was 36.8% (95% confidence interval [CI], 16.3–61.6) with 1 complete response and 6 partial responses. Confirmed PSA-RR was 46.3% (95% CI, 30.7–62.6). Median rPFS was 8.2 months (95% CI, 6.6–not estimable). The 6-month rPFS rate was 71.5%. Any-grade and grade 3/4 treatment-related adverse events occurred in 92.7% and 48.8% of pts, respectively. Associations of biomarkers are also evaluated. Conclusions The combination of NIVO+DOCE showed encouraging clinical activity in pts with mCRPC, with a safety profile consistent with those of the individual agents. A phase III trial is warranted to further evaluate NIVO+DOCE in mCRPC. Clinical trial identification NCT03338790. Editorial acknowledgement Professional medical writing assistance was provided by Nicolette Belletier, PhD, of Parexel, and funded by Bristol-Myers Squibb. Legal entity responsible for the study Bristol-Myers Squibb. Funding Bristol-Myers Squibb and Ono Pharmaceutical Company Limited. Disclosure K. Fizazi: Advisory / Consultancy: Janssen; Advisory / Consultancy: Astellas; Advisory / Consultancy: Bayer; Advisory / Consultancy: Orion. A. Rezazadeh Kalebasty: Advisory / Consultancy, Research grant / Funding (institution), Travel / Accommodation / Expenses: Genentech; Honoraria (self), Advisory / Consultancy, Speaker Bureau / Expert testimony, Research grant / Funding (institution), Travel / Accommodation / Expenses: Exelexis; Honoraria (self), Speaker Bureau / Expert testimony, Research grant / Funding (institution), Travel / Accommodation / Expenses: Janssen; Honoraria (self), Speaker Bureau / Expert testimony, Research grant / Funding (institution), Travel / Accommodation / Expenses: AZ; Advisory / Consultancy, Research grant / Funding (institution), Travel / Accommodation / Expenses: Bayer; Honoraria (self), Speaker Bureau / Expert testimony, Research grant / Funding (institution): BMS; Honoraria (self), Speaker Bureau / Expert testimony, Research grant / Funding (institution), Travel / Accommodation / Expenses: Eisai; Research grant / Funding (institution): Macrogenics; Honoraria (self), Speaker Bureau / Expert testimony, Research grant / Funding (institution), Travel / Accommodation / Expenses: Astellas; Research grant / Funding (institution): Beyond Spring; Research grant / Funding (institution): Bioclin/Rainier Therapeutics; Research grant / Funding (institution): Clovis; Research grant / Funding (institution): Bavarian; Research grant / Funding (institution): Nordic; Research grant / Funding (institution): Seattle Genetics; Honoraria (self), Advisory / Consultancy, Speaker Bureau / Expert testimony, Travel / Accommodation / Expenses: Novartis; Honoraria (self), Advisory / Consultancy, Speaker Bureau / Expert testimony, Travel / Accommodation / Expenses: Pfizer; Shareholder / Stockholder / Stock options: ECOM; Honoraria (self), Speaker Bureau / Expert testimony: Sanofi; Honoraria (self), Speaker Bureau / Expert testimony: Amgen; Travel / Accommodation / Expenses: Prometheus Labs. J.C. Vazquez Limon: Non-remunerated activity/ies, Personal Fees and non-financial support: AstraZeneca; Non-remunerated activity/ies, Non-financial support: Merck; Non-remunerated activity/ies, Personal fees: Roche; Non-remunerated activity/ies, Personal fees: Bristol-Myers Squibb. A.J. Armstrong: Research grant / Funding (institution): BMS; Advisory / Consultancy, Research grant / Funding (institution): Janssen; Advisory / Consultancy, Research grant / Funding (institution): Pfizer/Astellas; Advisory / Consultancy, Research grant / Funding (institution): Merck; Advisory / Consultancy, Speaker Bureau / Expert testimony, Research grant / Funding (institution): Dendreon; Advisory / Consultancy, Speaker Bureau / Expert testimony, Research grant / Funding (institution): Byer; Research grant / Funding (institution): Constellation; Advisory / Consultancy, Research grant / Funding (institution): AstraZeneca; Research grant / Funding (institution): Genentech/Roche; Advisory / Consultancy: Clovis. B. Sharkey: Full / Part-time employment: Bristol-Myers Squibb. A. Saci: Shareholder / Stockholder / Stock options, Full / Part-time employment: Bristol-Myers Squibb. J. Li: Shareholder / Stockholder / Stock options, Full / Part-time employment: Bristol-Myers Squibb. X. Wang: Full / Part-time employment: Bristol-Myers Squibb. M. Ciprotti: Full / Part-time employment: Bristol-Myers Squibb. P. Sathyanarayana: Shareholder / Stockholder / Stock options, Full / Part-time employment: Bristol-Myers Squibb. F. Saad: Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution): Astellas; Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution): Janssen; Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution): Bayer; Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution): Sanofi; Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution): BMS; Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution): AstraZeneca; Honoraria (self), Advisory / Consultancy, Research grant / Funding (institution): Pfizer. D.P. Petrylak: Advisory / Consultancy, Research grant / Funding (institution): AstraZeneca; Advisory / Consultancy, Research grant / Funding (institution): Bayer; Advisory / Consultancy, Shareholder / Stockholder / Stock options: Bellicum; Research grant / Funding (institution): Clovis; Advisory / Consultancy, Research grant / Funding (institution): Dendreon; Research grant / Funding (institution): Eli Lilly; Research grant / Funding (institution): Endocyte ; Advisory / Consultancy: Exelixis; Advisory / Consultancy: Ferring; Research grant / Funding (institution): Genentech; Research grant / Funding (institution): Innocrin; Advisory / Consultancy, Research grant / Funding (institution): Johnson and Johnson; Research grant / Funding (institution): MedImmune; Research grant / Funding (institution): Merck; Advisory / Consultancy, Research grant / Funding (institution): Millineum; Research grant / Funding (institution): Novartis; Advisory / Consultancy, Research grant / Funding (institution): Pfizer; Research grant / Funding (institution): Progenics; Research grant / Funding (institution): Roche Labatories; Advisory / Consultancy, Research grant / Funding (institution): Sanofi Aventis; Research grant / Funding (institution): Sotio; Speaker Bureau / Expert testimony, Research grant / Funding (institution), Shareholder / Stockholder / Stock options: Tyme Pharmaceuticals. R.K. Pachynski: Advisory / Consultancy, Research grant / Funding (institution): BMS; Speaker Bureau / Expert testimony: Dendreon; Advisory / Consultancy: EMD Serono/Pfizer; Speaker Bureau / Expert testimony, Travel / Accommodation / Expenses: Genentech/Roche; Advisory / Consultancy, Speaker Bureau / Expert testimony: Genomic Health; Research grant / Funding (institution): Janssen; Advisory / Consultancy, Speaker Bureau / Expert testimony: AstraZeneca; Advisory / Consultancy, Speaker Bureau / Expert testimony: Sanofi; Speaker Bureau / Expert testimony: Merck; Advisory / Consultancy: Jounce Therapeutics. C. Drake: Advisory / Consultancy, Research grant / Funding (institution), Shareholder / Stockholder / Stock options: Compugen; Advisory / Consultancy: Roche/Genentech; Advisory / Consultancy, Research grant / Funding (institution): Janssen Oncology; Advisory / Consultancy: Pfizer; Advisory / Consultancy: Astellas Medivation; Advisory / Consultancy, Shareholder / Stockholder / Stock options: Tizona Therapeutics; Advisory / Consultancy: Merck; Advisory / Consultancy: Pierre Fabre; Advisory / Consultancy: F-Star; Advisory / Consultancy, Shareholder / Stockholder / Stock options: KLEO; Advisory / Consultancy, Shareholder / Stockholder / Stock options: Harpoon; Advisory / Consultancy, Shareholder / Stockholder / Stock options: Shattuck Labs; Advisory / Consultancy, Shareholder / Stockholder / Stock options: Werewolf; Advisory / Consultancy: Genocea; Advisory / Consultancy: Sanofi Aventis; Advisory / Consultancy, Licensing / Royalties: Amplimmune; Advisory / Consultancy, Licensing / Royalties: BMS; Advisory / Consultancy: EMD. All other authors have declared no conflicts of interest.

In content-based image retrieval, feature representation is a fundamental issue and the selection of features significantly impacts the retrieval performance. Generally, the performance of a single visual feature is limited and the fusion with multiple complementary features is a preferable solution to boost the accuracy of image retrieval. In this paper, we propose an adaptively weighted graph fusion method to boost the retrieval accuracy. Given a query image, for each feature, one graph is constructed based on the nearest-neighborhood of database images. In order to evaluate the quality of this feature, we firstly use the PageRank algorithm to analyse the linked graph and get the corresponding PageRank value of each relevant image, which can be regarded as a conditional probability. Then, the weight of each feature is determined by the distribution of the above probability. We evaluate our method on two public datasets and the experimental results demonstrate that our method improves the retrieval performance consistently.

A 1:8 foundation model was designed and fabricated based on a spring vibration isolation steam-turbine generator foundation. The gravity was simulated through the hydraulic servo loading system, and the pseudo dynamic tests were carried out under the action of two different 7th intensity medium earthquakes and a 8th intensity rare earthquake. The results indicate that, under the action of different earthquakes, the acceleration amplification factors are all smaller, the inter-story drift angles of each story are in accordance with the specification requirements. Cracks appeared under the action of 8th intensity rare earthquake , but no penetrating cracks. Springs have played a good role in vibration isolation, and no relative slippages between the spring vibration isolator and the capital of columns.

Nature Communications 8: Article number: 14864 (2017)); Published: 28 March 2017; Updated: 25 April 2017 The original version of this Article contained a typographical error in the spelling of the author Beifang Niu, which was incorrectly given as Beifung Niu. This has now been corrected in both thePDF and HTML versions of the Article.

3579 Background: Dasatinib is a potent, multi-targeted kinase inhibitor that was recently approved for treatment of chronic myelogenous leukemia resistant to imatinib. To aid its clinical development in prostate cancer, we used a panel of prostate cancer cell lines to identify molecular markers that could be used to predict sensitivity to dasatinib and to monitor its activity. Methods: Baseline gene expression profiles of 16 cell lines were used to identify predictive biomarkers based on the correlation of gene expression with in vitro sensitivity of cells to dasatinib. Selected cell lines were treated with dasatinib to identify surrogate biomarkers based upon changes in gene expression following dasatinib treatment. Results: We identified 174 genes whose baseline expression levels were highly correlated with sensitivity or resistance to dasatinib. These include cell lineage markers cytokeratin 5 (CK5), androgen receptor (AR), and prostate specific antigen (PSA). Our results indicate that “basal type” cell lines (those with high expression of CK5 and low expression of AR and PSA) are sensitive to dasatinib. Dasatinib treatment studies further identified genes whose expression levels were significantly modulated by the drug. Ten genes, including urokinase-type plasminogen activator (uPA), were not only significantly correlated with sensitivity to dasatinib but also reduced in their expression upon drug treatment. In addition, the down-regulation of uPA was specific to dasatinib and the effect was not seen in taxol-treated cells. The extent of down-regulation was correlated with the sensitivity of cell lines to dasatinib. EphA2, a specific kinase target of dasatinib, was identified as a biomarker common to prostate and breast cancers. Finally, the expression of 5 genes including CK5, AR, PSA, uPA and EphA2 in prostate tumors was examined and the dasatinib sensitivity signature was validated using a published data set derived from a clinical population. Conclusions: Candidate markers correlated with dasatinib sensitivity were identified. A five-gene model consisting of predictive markers as well as potential surrogate markers has been formulated and will be evaluated in ongoing dasatinib prostate cancer trials. No significant financial relationships to disclose.