Search

Advanced high-grade serous ovarian cancer (HGSC) is an aggressive disease that accounts for 70% of all ovarian cancer deaths. Nevertheless, 15% of patients diagnosed with advanced HGSC survive more than 10 years. The elucidation of predictive markers of these long-term survivors (LTS) could help identify therapeutic targets for the disease, and thus improve patient survival rates. To investigate the stromal heterogeneity of the tumor microenvironment (TME) in ovarian cancer, we used spatial transcriptomics to generate spatially resolved transcript profiles in treatment-naïve advanced HGSC from LTS and short-term survivors (STS) and determined the association between cancer-associated fibroblasts (CAF) heterogeneity and survival in patients with advanced HGSC. Spatial transcriptomics and single-cell RNA-sequencing data were integrated to distinguish tumor and stroma regions, and a computational method was developed to investigate spatially resolved ligand–receptor interactions between various tumor and CAF subtypes in the TME. A specific subtype of CAFs and its spatial location relative to a particular ovarian cancer cell subtype in the TME correlated with long-term survival in patients with advanced HGSC. Also, increased APOE-LRP5 cross-talk occurred at the stroma-tumor interface in tumor tissues from STS compared with LTS. These findings were validated using multiplex IHC. Overall, this spatial transcriptomics analysis revealed spatially resolved CAF-tumor cross-talk signaling networks in the ovarian TME that are associated with long-term survival of patients with HGSC. Further studies to confirm whether such cross-talk plays a role in modulating the malignant phenotype of HGSC and could serve as a predictive biomarker of patient survival are warranted. Significance: Generation of spatially resolved gene expression patterns in tumors from patients with ovarian cancer surviving more than 10 years allows the identification of novel predictive biomarkers and therapeutic targets for better patient management. See related commentary by Kelliher and Lengyel, p. 1383

Supplementary Figures 1-11

Supplementary Materials and Methods

Spatial Transcriptomics differentially expressed genes

MIA immune enrichment group comparison

scRNAseq differentially expressed genes

Advanced high-grade serous ovarian cancer (HGSC) is an aggressive disease that accounts for 70% of all ovarian cancer deaths. Nevertheless, 15% of patients diagnosed with advanced HGSC survive more than 10 years. The elucidation of predictive markers of these long-term survivors (LTS) could help identify therapeutic targets for the disease, and thus improve patient survival rates. To investigate the stromal heterogeneity of the tumor microenvironment (TME) in ovarian cancer, we used spatial transcriptomics to generate spatially resolved transcript profiles in treatment-naïve advanced HGSC from LTS and short-term survivors (STS) and determined the association between cancer-associated fibroblasts (CAF) heterogeneity and survival in patients with advanced HGSC. Spatial transcriptomics and single-cell RNA-sequencing data were integrated to distinguish tumor and stroma regions, and a computational method was developed to investigate spatially resolved ligand–receptor interactions between various tumor and CAF subtypes in the TME. A specific subtype of CAFs and its spatial location relative to a particular ovarian cancer cell subtype in the TME correlated with long-term survival in patients with advanced HGSC. Also, increased APOE-LRP5 cross-talk occurred at the stroma-tumor interface in tumor tissues from STS compared with LTS. These findings were validated using multiplex IHC. Overall, this spatial transcriptomics analysis revealed spatially resolved CAF-tumor cross-talk signaling networks in the ovarian TME that are associated with long-term survival of patients with HGSC. Further studies to confirm whether such cross-talk plays a role in modulating the malignant phenotype of HGSC and could serve as a predictive biomarker of patient survival are warranted.Significance:Generation of spatially resolved gene expression patterns in tumors from patients with ovarian cancer surviving more than 10 years allows the identification of novel predictive biomarkers and therapeutic targets for better patient management.See related commentary by Kelliher and Lengyel, p. 1383

Clinicopathological characteristics of samples used in the study

Supplemantary Video 3, LLC1 cells cocultured with NG2- BMSCs

summarized excel file containing the quantitative raw data and processed values in the manuscript

Supplemantary Video 4, LLC1 cells cocultured with NG2+ Ncadherin KO BMSCs

The bone microenvironment is dynamic and undergoes remodeling in normal and pathologic conditions. Whether such remodeling affects disseminated tumor cells (DTC) and bone metastasis remains poorly understood. Here, we demonstrated that pathologic fractures increase metastatic colonization around the injury. NG2+ cells are a common participant in bone metastasis initiation and bone remodeling in both homeostatic and fractured conditions. NG2+ bone mesenchymal stem/stromal cells (BMSC) often colocalize with DTCs in the perivascular niche. Both DTCs and NG2+ BMSCs are recruited to remodeling sites. Ablation of NG2+ lineage impaired bone remodeling and concurrently diminished metastatic colonization. In cocultures, NG2+ BMSCs, especially when undergoing osteodifferentiation, enhanced cancer cell proliferation and migration. Knockout of N-cadherin in NG2+ cells abolished these effects in vitro and phenocopied NG2+ lineage depletion in vivo. These findings uncover dual roles of NG2+ cells in metastasis and remodeling and indicate that osteodifferentiation of BMSCs promotes metastasis initiation via N-cadherin–mediated cell–cell interaction.Significance:The bone colonization of cancer cells occurs in an environment that undergoes constant remodeling. Our study provides mechanistic insights into how bone homeostasis and pathologic repair lead to the outgrowth of disseminated cancer cells, thereby opening new directions for further etiologic and epidemiologic studies of tumor recurrences.This article is highlighted in the In This Issue feature, p. 247

Supplemantary Video 1, LLC1 cells cocultured with NG2+ BMSCs

Supplementary figures and corresponding figure legends

Mounting clinical and preclinical evidence supports a key role for sustained adrenergic signaling in the tumor microenvironment as a driver of tumor growth and progression. However, the mechanisms by which adrenergic neurotransmitters are delivered to the tumor microenvironment are not well understood. Here we present evidence for a feed-forward loop whereby adrenergic signaling leads to increased tumoral innervation. In response to catecholamines, tumor cells produced brain-derived neurotrophic factor (BDNF) in an ADRB3/cAMP/Epac/JNK-dependent manner. Elevated BDNF levels in the tumor microenvironment increased innervation by signaling through host neurotrophic receptor tyrosine kinase 2 receptors. In patients with cancer, high tumor nerve counts were significantly associated with increased BDNF and norepinephrine levels and decreased overall survival. Collectively, these data describe a novel pathway for tumor innervation, with resultant biological and clinical implications.Significance: Sustained adrenergic signaling promotes tumor growth and metastasis through BDNF-mediated tumoral innervation. Cancer Res; 78(12); 3233–42. ©2018 AACR.

The activation of the epithelial-to-mesenchymal transition (EMT) program is a critical step in cancer progression and metastasis, but visualization of this process at the single-cell level, especially in vivo, remains challenging. We established an in vivo approach to track the fate of tumor cells based on a novel EMT-driven fluorescent color switching breast cancer mouse model and intravital two-photon laser scanning microscopy. Specifically, the MMTV-PyMT, Rosa26-RFP-GFP, and Fsp1-Cre triple transgenic mouse model was used to monitor the conversion of RFP-positive epithelial cells to GFP-positive mesenchymal cells in mammary tumors under the control of the Fsp1 (ATL1) promoter, a gate-keeper of EMT initiation. RFP-positive cells were isolated from the tumors, sorted, and transplanted into mammary fat pads of SCID mice to monitor EMT during breast tumor formation. We found that the conversion from RFP- to GFP-positive and spindle-shaped cells was a gradual process, and that GFP-positive cells preferentially localized close to blood vessels, independent of tumor size. Furthermore, cells undergoing EMT expressed high levels of the HGF receptor, c-Met, and treatment of RFP-positive cells with the c-Met inhibitor, cabozantinib, suppressed the RFP-to-GFP conversion in vitro. Moreover, administration of cabozantinib to mice with palpable RFP-positive tumors resulted in a silent EMT phenotype whereby GFP-positive cells exhibited reduced motility, leading to suppressed tumor growth. In conclusion, our imaging technique provides a novel opportunity for visualizing tumor EMT at the single-cell level and may help to reveal the intricacies underlying tumor dynamics and treatment responses. Cancer Res; 76(8); 2094–104. ©2016 AACR.

Purpose:Recent studies demonstrate the role of the tumor microenvironment in tumor progression. However, strategies used to overcome the malignant phenotypes of cancer cells modulated by the microenvironment have not been thoroughly explored. In this study, we evaluated the therapeutic efficacy of a newly developed mAb targeting microfibril-associated protein 5 (MFAP5), which is secreted predominately by cancer-associated fibroblast (CAF), in ovarian and pancreatic cancer models.Experimental Design: MAbs were developed using human MFAP5 recombinant protein as an antigen in mice, and antibodies from hybridoma clones were evaluated for their specificity to human and murine MFAP5. An Octet RED384 system was used to determine the kinetics of binding affinity and the specificity of the antibody clones, which were followed by epitope mapping and functional characterization by in vitro assays. The therapeutic efficacy of a lead anti-MFAP5 antibody clone 130A in tumor suppression was evaluated by ovarian tumor– and pancreatic tumor–bearing mouse models.Results:Three hybridoma clones, which produced antibodies with high affinity and specificity to MFAP5, were selected for functional studies. Antibody clone 130A, which recognizes a common epitope shared between human and murine MFAP5 protein, was further selected for in vivo studies. Results showed that clone 130A downregulated MFAP5-induced collagen production in CAFs, suppressed intratumoral microvessel leakiness, and enhanced paclitaxel bioavailability in both ovarian and pancreatic cancer mouse models.Conclusions:These data suggest that MFAP5 blockade using an immunologic approach inhibits fibrosis, induces tumor vessel normalization, and enhances chemosensitivity in ovarian and pancreatic cancer, and can be used as a novel therapeutic agent.

Purpose: Transforming growth factor β1 (Tgfβ1) plays an important role in cancer. Most of Tgfβ1 in plasma is from platelets; thus, we studied whether platelet Tgfβ1 has any role in the progression of ovarian cancer, and whether this role is limited to metastasis or also involves the growth of primary tumors.Experimental Design: We compared the growth of murine ovarian cancer cell-induced tumors in platelet-specific Tgfβ1-deficient mice and wild-type mice. Using resected tumor nodules, we studied the effect of platelet Tgfβ1 on neoangiogenesis and on platelet extravasation into tumors. To investigate the effect of Tgfβ1 at different stages of ovarian cancer, we reduced expression of Tgfβ1 receptor (its TgfβR1 component) in tumors at different time points after injection of cancer cells, and compared the final tumor size.Results: Lack of platelet Tgfβ1 in mice reduced tumor growth, neoangiogenesis, and platelet extravasation. Ovarian cancer tumors in platelet-specific Tgfβ1-deficient mice reached less than half of their size in wild-type littermates. Knockdown of TgfβR1 on cancer cells in the first 2 weeks after their injection reduced tumor growth, but was less effective if initiated after 3 weeks.Conclusions: We showed that platelet Tgfβ1 increased the growth of primary tumors in murine models of ovarian cancer. We also showed that inhibition of TgfβR1 is more effective in reducing the growth of ovarian cancer if initiated earlier. Our results supported a therapeutic benefit in preventing platelet activation, degranulation, and release of Tgfβ1 in ovarian cancer. Clin Cancer Res; 23(18); 5611–21. ©2017 AACR.

Supplementary figures and methods

Figure S1: Effect of platelet Tgf�1 and ovarian cancer cell Tgf�R1 on the number of tumor nodules in orthotopic murine models of ovarian cancer. Figure S2: Quantification of Tgf�R1 mRNA level by qRT-PCR in IG10 and ID8 murine ovarian cancer cells incubated with scrambled siRNA or human specific hTgf�R1 siRNA in vitro for 48 hours (n=6). Figure S3: Effect of platelet Tgf�1 on proliferation of human ovarian cancer cells in vitro. Platelets isolated from whole blood of mice were coincubated with human ovarian cancer cells (SKOV3ip1, OVCAR432, and CaOV3) and cell proliferation rate was measured by quantifying BrdU incorporations

Supplementary figure 1: Scheme of the imaging setup and representative in vivo images Supplementary figure 2: Representative in vivo images of GFP+ cells in middle and large tumors Supplementary figure 3: High expression of c-Met and Flt 1/4 correlate with metastasis-free survival in breast cancer patients

Table 1: Distribution of the GFP+ cells in whole tumor central vs peripheral area and tumor lobule central vs peripheral area. Table 2: RNAseq gene expressions of growth factor receptors in sorted RFP+ and GFP+ cells.

A representative video showing the Z�stack IVM images of a 2mm�diameter pure RFP+ cellinjected tumor in the wild type mouse. Sporadic GFP+ cells with cuboidal morphology were examined in the tumor.

PDF file - 506KB, Supp. Fig. S1 2D structures of Amoxapine, Amoxapine's metabolites, Loxapine, and the previous reported potent GUS inhibitors - Inhibitor 1 and 2. Suppl. Fig. S2 The initial structures (left panel, Before MD) and the last snapshots (right panel, After MD) taken from 4 ns MD simulation. Supp. Table S1. Hydrogen bonds formed by ligand and protein during the last 2 ns simulation. Supp. Table. S2 Histologic score and Ki-67-positive cells of the examined tissue.

CARS imaging of ex vivo ovarian cancer samples

Fig. S1. Sustained adrenergic signaling increases nerve counts in tumors. Fig. S2. Characterization of tumoral innervation. Fig. S3. NE induces BDNF expression. Fig. S4. NE-induced BDNF expression is mediated by ADRB3/Epac/Jnk. Fig. S5. BDNF increases nerve counts. Fig. S6. Adrenergic-mediated mTrkB activation leads to increased in vivo tumor nodule counts. Table S1. Alteration in pathways associated with neuronal growth and function after NE treatment (HeyA8 and SKOV3ip1 cells). Table S2. Association of Clinicopathologic variables with BDNF protein expression. Table S3. Association of Clinicopathologic variables with Nerve Counts.

supplementary figure legends

Purpose: Irinotecan (CPT-11) induced diarrhea occurs frequently in patients with cancer and limits its usage. Bacteria β-glucuronidase (GUS) enzymes in intestines convert the nontoxic metabolite of CPT-11, SN-38G, to toxic SN-38, and finally lead to damage of intestinal epithelial cells and diarrhea. We previously reported amoxapine as a potent GUS inhibitor in vitro. To further understand the molecular mechanism of amoxapine and its potential for treatment of CPT-11–induced diarrhea, we studied the binding modes of amoxapine and its metabolites by docking and molecular dynamics simulation, and tested the in vivo efficacy on mice in combination with CPT-11.Experimental Design: The binding of amoxapine, its metabolites, 7-hydroxyamoxapine and 8-hydroxyamoxapine, and a control drug loxapine with GUS was explored by computational protocols. The in vitro potencies of metabolites were measured by Escherichia coli GUS enzyme and cell-based assay. Low-dosage daily oral administration was designed to use along with CPT-11 to treat tumor-bearing mice.Results: Computational modeling results indicated that amoxapine and its metabolites bound in the active site of GUS and satisfied critical pharmacophore features: aromatic features near bacterial loop residue F365′ and hydrogen bond toward E413. Amoxapine and its metabolites were demonstrated as potent in vitro. Administration of low dosages of amoxapine with CPT-11 in mice achieved significant suppression of diarrhea and reduced tumor growth.Conclusions: Amoxapine has great clinical potential to be rapidly translated to human subjects for irinotecan-induced diarrhea. Clin Cancer Res; 20(13); 3521–30. ©2014 AACR.

CARS imaging of ex vivo ovarian cancer samples

Cancer cells with active drug efflux capability are multidrug resistant and pose a significant obstacle for the efficacy of chemotherapy. Moreover, recent evidence suggests that high drug efflux cancer cells (HDECC) may be selectively enriched with stem-like cancer cells, which are believed to be the cause for tumor initiation and recurrence. There is a great need for therapeutic reagents that are capable of eliminating HDECCs. We developed an image-based high-content screening (HCS) system to specifically identify and analyze the HDECC population in lung cancer cells. Using the system, we screened 1,280 pharmacologically active compounds that identified 12 potent HDECC inhibitors. It is shown that these inhibitors are able to overcome multidrug resistance (MDR) and sensitize HDECCs to chemotherapeutic drugs, or directly reduce the tumorigenicity of lung cancer cells possibly by affecting stem-like cancer cells. The HCS system we established provides a new approach for identifying therapeutic reagents overcoming MDR. The compounds identified by the screening may potentially be used as potential adjuvant to improve the efficacy of chemotherapeutic drugs. Cancer Res; 70(19); 7723–33. ©2010 AACR.

XLS file - 7MB

Little research has been done to address the huge opportunities that may exist to reposition existing approved or generic drugs for alternate uses in cancer therapy. In addition, there has been little work on strategies to reposition experimental cancer agents for testing in alternate settings that could shorten their clinical development time. Progress in each area has lagged, in part, because of the lack of systematic methods to define drug off-target effects (OTE) that might affect important cancer cell signaling pathways. In this study, we addressed this critical gap by developing an OTE-based method to repurpose drugs for cancer therapeutics, based on transcriptional responses made in cells before and after drug treatment. Specifically, we defined a new network component called cancer-signaling bridges (CSB) and integrated it with a Bayesian factor regression model (BFRM) to form a new hybrid method termed CSB-BFRM. Proof-of-concept studies were conducted in breast and prostate cancer cells and in promyelocytic leukemia cells. In each system, CSB-BFRM analysis could accurately predict clinical responses to more than 90% of drugs approved by the U.S. Food and Drug Administration and more than 75% of experimental clinical drugs that were tested. Mechanistic investigation of OTEs for several high-ranking drug–dose pairs suggested repositioning opportunities for cancer therapy, based on the ability to enforce retinoblastoma-dependent repression of important E2F-dependent cell-cycle genes. Together, our findings establish new methods to identify opportunities for drug repositioning or to elucidate the mechanisms of action of repositioned drugs. Cancer Res; 72(1); 33–44. ©2011 AACR.

PDf file - 3.4MB

Alzheimer’s disease and related dementias (AD/ADRD) affects more than 50 million people worldwide but there is no clear therapeutic option affordable for the general patient population. Recently, drug repositioning studies featuring collaborations between academic institutes, medical centers, and hospitals are generating novel therapeutics candidates against these devastating diseases and filling in an important area for healthcare that is poorly represented by pharmaceutical companies. Such drug repositioning studies converge expertise from bioinformatics, chemical informatics, medical informatics, artificial intelligence, high throughput and high-content screening and systems biology. They also take advantage of multi-scale, multi-modality datasets, ranging from transcriptomic and proteomic data, electronical medical records, and medical imaging to social media information of patient behaviors and emotions and epidemiology profiles of disease populations, in order to gain comprehensive understanding of disease mechanisms and drug effects. We proposed a recursive drug repositioning paradigm involving the iteration of three processing steps of modeling, prediction, and validation to identify known drugs and bioactive compounds for AD/ADRD. This recursive paradigm has the potential of quickly obtaining a panel of robust novel drug candidates for AD/ADRD and gaining in-depth understanding of disease mechanisms from those repositioned drug candidates, subsequently improving the success rate of predicting novel hits.

The bone microenvironment is dynamic and undergoes remodeling in normal and pathologic conditions. Whether such remodeling affects disseminated tumor cells (DTC) and bone metastasis remains poorly understood. Here, we demonstrated that pathologic fractures increase metastatic colonization around the injury. NG2+ cells are a common participant in bone metastasis initiation and bone remodeling in both homeostatic and fractured conditions. NG2+ bone mesenchymal stem/stromal cells (BMSC) often colocalize with DTCs in the perivascular niche. Both DTCs and NG2+ BMSCs are recruited to remodeling sites. Ablation of NG2+ lineage impaired bone remodeling and concurrently diminished metastatic colonization. In cocultures, NG2+ BMSCs, especially when undergoing osteodifferentiation, enhanced cancer cell proliferation and migration. Knockout of N-cadherin in NG2+ cells abolished these effects in vitro and phenocopied NG2+ lineage depletion in vivo. These findings uncover dual roles of NG2+ cells in metastasis and remodeling and indicate that osteodifferentiation of BMSCs promotes metastasis initiation via N-cadherin–mediated cell–cell interaction. Significance: The bone colonization of cancer cells occurs in an environment that undergoes constant remodeling. Our study provides mechanistic insights into how bone homeostasis and pathologic repair lead to the outgrowth of disseminated cancer cells, thereby opening new directions for further etiologic and epidemiologic studies of tumor recurrences. This article is highlighted in the In This Issue feature, p. 247

Advanced high-grade serous ovarian cancer (HGSC) is an aggressive disease that accounts for 70% of all ovarian cancer deaths. Nevertheless, 15% of patients diagnosed with advanced HGSC survive more than 10 years. The identification of predictive markers associated with tumors developed from these long-term survivors (LTS) is crucial to identifying therapeutic targets for the disease, and thus improving patient survival rates. Reports to date have not fully established the stromal heterogeneity of the tumor microenvironment (TME) in ovarian cancer and its association with clinical outcomes. We used a spatial transcriptomics platform to generate spatially resolved transcript profiles in treatment naïve advanced HGSC from LTS and short-term survivors (STS), and determined whether cancer-associated fibroblasts (CAFs) heterogeneity is associated with survival in patients with advanced HGSC. We integrated spatial transcriptomics with single-cell RNA sequencing data to distinguish tumor and stroma regions, and developed a method to investigate spatially resolved ligand-receptor interactions between various tumor and CAF subtypes in the TME. In addition, we used multiplex immunohistochemistry techniques to validate our findings. We found that a specific subtype of CAFs and its spatial location relative to a particular ovarian cancer cell subtype in the TME correlate with long-term survival in advanced HGSC patients. We also demonstrated that significant APOE-LRP5 crosstalk occurred at the stroma-tumor interface in tumor tissues from STS compared to LTS, suggesting that such crosstalk plays a crucial role in modulating the malignant phenotype of HGSC, and could serve as a predictive biomarker of patient survival.

Background: Stroke infarct volume predicts patient disability and has utility for clinical trial outcomes. Accurate infarct volume measurement requires manual segmentation of stroke boundaries in diffusion-weighted magnetic resonance imaging scans which is time-consuming and subject to variability. Automatic infarct segmentation should be robust to rotation and reflection; however, prior work has not encoded this property into deep learning architecture. Here, we use rotation-reflection equivariance and train a deep learning model to segment stroke volumes in a large cohort of well-characterized patients with acute ischemic stroke in different vascular territories. Methods: In this retrospective study, patients were selected from a stroke registry at Houston Methodist Hospital. Eight hundred seventy-five patients with acute ischemic stroke in any brain area who had magnetic resonance imaging with diffusion-weighted imaging were included for analysis and split 80/20 for training/testing. Infarct volumes were manually segmented by consensus of 3 independent clinical experts and cross-referenced against radiology reports. A rotation-reflection equivariant model was developed based on U-Net and grouped convolutions. Segmentation performance was evaluated using Dice score, precision, and recall. Ninety-day modified Rankin Scale outcome prediction was also evaluated using clinical variables and segmented stroke volumes in different brain regions. Results: Segmentation model Dice scores are 0.88 (95% CI, 0.87–0.89; training) and 0.85 (0.82–0.88; testing). The modified Rankin Scale outcome prediction AUC using stroke volume in 30 refined brain regions based upon modified Rankin Scale-relevance areas adjusted for clinical variables was 0.80 (0.76–0.83) with an accuracy of 0.75 (0.72–0.78). Conclusions: We trained a deep learning model with encoded rotation-reflection equivariance to segment acute ischemic stroke lesions in diffusion- weighted imaging using a large data set from the Houston Methodist stroke center. The model achieved competitive performance in 175 well-balanced hold-out testing cases that include strokes from different vascular territories. Furthermore, the location specific stroke volume segmentations from the deep learning model combined with clinical factors demonstrated high AUC and accuracy for 90-day modified Rankin Scale in an outcome prediction model.

Probability of malignancy for BI-RADS 4-designated breast lesions ranges from 2% to 95%, contributing to high false-positive biopsy rates. We compare clinical performance of digital breast tomosynthesis (DBT) versus digital mammography (2D) among our BI-RADS 4 population without prior history of breast cancer.We extracted retrospective data i.e., clinical, mammogram reports, and biopsy data, from electronic medical records across Houston Methodist's nine hospitals for patients who underwent diagnostic examinations using both modalities (02/01/2015 - 09/30/2020). 2D and DBT cohorts were not intra-individual matched, and there was no direct mammogram evaluation. Using Student's t test, Fisher's exact test, and Chi-squared test, we evaluated the data to determine statistical significance of differences between modalities in BI-RADS 4 cases. We calculated adjusted odds-ratio between modalities for cancer detection rate (CDR) and biopsy-derived positive predictive value (PPV3).There were 6,356 encounters (6,020 patients) in 2D and 5,896 encounters (5,637 patients) in DBT assessed as BI-RADS 4. Using Fisher's exact test, DBT mammography cases were significantly assessed as BI-RADS 4 5.66% more often than those undergoing 2D mammography, P = 0.0046 (1.0566 95% CI: 1.0169-1.0977). The CDRs were 112.65 (2D) and 120.76 (DBT), adjusted odds-ratio: 1.04 (0.93, 1.16)), P = 0.5029, while PPV3 were 14.41% (2D) and 15.99% (DBT), adjusted odds-ratio: 1.09 (0.97, 1.22), P = 0.1483; both logistic regression-adjusted for all other factors.DBT did not achieve better performance and sensitivity in assigning BI-RADS 4 cases compared with 2D, showed no significant advantage in CDR and PPV3, and does not reduce false-positive biopsies among BI-RADS 4-assessed patients.

Epithelial to mesenchymal transition (EMT) plays an important role in metastasis and chemoresistance. However, targeting EMT has been challenging due to the complex and redundant EMT signaling pathways. Moreover, EMT inhibition would unequivocally promote its reversed process, mesenchymal to epithelial transition (MET), which also favors malignancy progression. To solve this EMT targeting paradox, we proposed to target the ability of tumor cells to oscillate between epithelial and mesenchymal states, the epithelial to mesenchymal plasticity (EMP), which encompasses both EMT and MET. Using a SnailCre-ERT2 mediated EMT lineage-tracing model, we confirmed the involvement of EMP in chemoresistant metastasis of breast cancer. Single-cell transcriptome analyses of tumor cells in the transitioning phases of EMT or MET consistently revealed that ribosome biogenesis (RiBi) was specifically elevated regardless of their destined phenotypical fates. Further analyses delineated that the upregulation of RiBi and the subsequent nascent protein synthesis via the intensified ERK and mTOR signalings were vital for cancer cells to transit between epithelial and mesenchymal states. Importantly, inhibiting the excessive RiBi by either genetic or pharmacological means impaired EMP, as both EMT and MET processes were impeded, leading to significantly reduced metastatic outgrowth of chemoresistant tumor cells. Our studies revealed that elevated RiBi is required for tumor cells to “Ping Pong” between epithelial and mesenchymal states. Targeting the EMP via RiBi inhibition provides a feasible strategy to improve therapies for advanced breast cancers. Citation Format: Yi Ban, Wenjuan Dong, Yue Zou, Sharrell B. Lee, Jianting Sheng, Yuliang Cao, Stephen T.C. Wong, Dingcheng Gao. Targeting epithelial-mesenchymal plasticity via ribosome biogenesis inhibition reduces chemoresistant metastasis of breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 963.

In an emergency room (ER) setting, stroke triage or screening is a common challenge. A quick CT is usually done instead of MRI due to MRI's slow throughput and high cost. Clinical tests are commonly referred to during the process, but the misdiagnosis rate remains high. We propose a novel multimodal deep learning framework, DeepStroke, to achieve computer-aided stroke presence assessment by recognizing patterns of minor facial muscles incoordination and speech inability for patients with suspicion of stroke in an acute setting. Our proposed DeepStroke takes one-minute facial video data and audio data readily available during stroke triage for local facial paralysis detection and global speech disorder analysis. Transfer learning was adopted to reduce face-attribute biases and improve generalizability. We leverage a multi-modal lateral fusion to combine the low- and high-level features and provide mutual regularization for joint training. Novel adversarial training is introduced to obtain identity-free and stroke-discriminative features. Experiments on our video-audio dataset with actual ER patients show that DeepStroke outperforms state-of-the-art models and achieves better performance than both a triage team and ER doctors, attaining a 10.94% higher sensitivity and maintaining 7.37% higher accuracy than traditional stroke triage when specificity is aligned. Meanwhile, each assessment can be completed in less than six minutes, demonstrating the framework's great potential for clinical translation.