Deep learning-enabled pelvic ultrasound images for accurate diagnosis of ovarian cancer in China: a retrospective, multicentre, diagnostic study.

BACKGROUND: Ultrasound is a critical non-invasive test for preoperative diagnosis of ovarian cancer. Deep learning is making advances in image-recognition tasks; therefore, we aimed to develop a deep convolutional neural network (DCNN) model that automates evaluation of ultrasound images and to facilitate a more accurate diagnosis of ovarian cancer than existing methods. METHODS: In this retrospective, multicentre, diagnostic study, we collected pelvic ultrasound images from ten hospitals across China between September 2003, and May 2019. We included consecutive adult patients (aged ≥18 years) with adnexal lesions in ultrasonography and healthy controls and excluded duplicated cases and patients without adnexa or pathological diagnosis. For DCNN model development, patients were assigned to the training dataset (34 488 images of 3755 patients with ovarian cancer, 541 442 images of 101 777 controls). For model validation, patients were assigned to the internal validation dataset (3031 images of 266 patients with ovarian cancer, 5385 images of 602 with benign adnexal lesions), external validation datasets 1 (486 images of 67 with ovarian cancer, 933 images of 268 with benign adnexal lesions), and 2 (1253 images of 166 with ovarian cancer, 5257 images of 723 benign adnexal lesions). Using these datasets, we assessed the diagnostic value of DCNN, compared DCNN with 35 radiologists, and explored whether DCNN could augment the diagnostic accuracy of six radiologists. Pathological diagnosis was the reference standard. FINDINGS: For DCNN to detect ovarian cancer, AUC was 0·911 (95% CI 0·886-0·936) in the internal dataset, 0·870 (95% CI 0·822-0·918) in external validation dataset 1, and 0·831 (95% CI 0·793-0·869) in external validation dataset 2. The DCNN model was more accurate than radiologists at detecting ovarian cancer in the internal dataset (88·8% vs 85·7%) and external validation dataset 1 (86·9% vs 81·1%). Accuracy and sensitivity of diagnosis increased more after DCNN-assisted diagnosis than assessment by radiologists alone (87·6% [85·0-90·2] vs 78·3% [72·1-84·5], p<0·0001; 82·7% [78·5-86·9] vs 70·4% [59·1-81·7], p<0·0001). The average accuracy of DCNN-assisted evaluations for six radiologists reached 0·876 and were significantly augmented when they were DCNN-assisted (p<0·05). INTERPRETATION: The performance of DCNN-enabled ultrasound exceeded the average diagnostic level of radiologists matched the level of expert ultrasound image readers, and augmented radiologists' accuracy. However, these observations warrant further investigations in prospective studies or randomised clinical trials. FUNDING: National Key Basic Research Program of China, National Sci-Tech Support Projects, and National Natural Science Foundation of China.

Prognostic implications of tumour-infiltrating lymphocytes for recurrence in epithelial ovarian cancer.

The recurrence of patients with epithelial ovarian cancer (EOC) is largely attributed to tumour cells escaping from the surveillance of immune cells. However, to date there is a lack of studies that have systematically evaluated the associations between the infiltration fraction of immune cells and the recurrence risk of EOC. Based on the micro-ribonucleic acid (microRNA) expression profiles of 441 EOC patients, we constructed a microRNA-based panel with recurrence prediction potential using non-negative matrix factorization consensus clustering. Then, we evaluated the association between recurrence risk and infiltration proportions among 10 immune cell types by CIBERSORT and a multivariable Cox regression model. As a result, we identified a 72-microRNA-based panel that could stratify patients into high and low risk of recurrence. The infiltration of plasma cells and M1 macrophages was consistently significantly associated with the risk of recurrence in patients with EOC. Plasma cells were significantly associated with a decreased risk of relapse [hazard ratio (HR) = 0.58, p = 0.006), while M1 macrophages were associated with an increased risk of relapse (HR = 1.59, p = 0.003). Therefore, the 72-microRNA-based panel, M1 macrophages and plasma cells may hold potential to serve as recurrence predictors of EOC patients in clinical practice.

The LIV-1-GRPEL1 axis adjusts cell fate during anti-mitotic agent-damaged mitosis.

BACKGROUND: Understanding how cells respond to mitotic poisons is of great biomedical and clinical significance. However, it remains unknown how cell-death or survival is determined during exposure to anti-mitotic drugs. METHODS: The biological effects of SLC39A6 (LIV-1) and GrpE-like 1 (GRPEL1) on mitotic exit and apoptosis were evaluated both in vitro and in vivo using flow cytometry, western blotting, xenografts and time-lapse imaging. The interactions between proteins and the ubiquitination of GRPEL1 were assessed by GST pull down, immunoprecipitation and mass spectrometry analysis. The expression of LIV-1 in cancers was assessed by immunohistochemistry. FINDINGS: Overexpression of LIV-1 led to direct apoptosis. Depleted for LIV-1 evade anti-mitotic agent-induced killing through a rapid exit from arrested mitosis. LIV-1 interacts with GRPEL1 and Stabilizes GRPEL1 Protein by Preventing Ubiquitylation of GRPEL1. LIV-1-GRPEL1 axis depletion works to reduce the mitotic arrest by inducing PP2A-B55α phosphates activity, while inhibit apoptosis by banding AIF and preventing the latter's release into the nucleus. Loss of function in this axis was frequent in multiple types of human epithelial cancer. INTERPRETATION: These data demonstrate that LIV-1-GRPEL1 axis dually regulates mitotic exit as well as apoptosis by interacting with PP2A B55α and AIF. Its discovery constitutes a conceptual advance for the decisive mechanism of cell fate during damaged mitosis. FUND: National Clinical Research Center for Obstetric and Gynecologic Diseases, the National Natural Science Foundation of China.

Correction: Loss of the novel mitochondrial protein FAM210B promotes metastasis via PDK4-dependent metabolic reprogramming.

Following publication of this article [1], the authors became aware of an error in Fig. 7e which requires correction. The images do not currently match the correct treatment and/or control conditions. Specifically, the images of siNC+AD-ctr (the top left panel) and siPDK4+AD-PDK4 (the bottom right panel) were incorrect. The error does not impact the conclusions of the article. They sincerely apologize for the mistakes in the article and any inconvenience caused.

Ubiquinol-cytochrome C reductase core protein II promotes tumorigenesis by facilitating p53 degradation.

BACKGROUND: Ubiquinol-cytochrome C reductase core protein II (QCR2) is essential for mitochondrial functions, yet, its role in cancer development has remained elusive. METHODS: The expression of QCR2 in cancer patients was assessed by immunohistochemistry. The proliferation of cancer cells was assessed by CCK-8 assay, EdU staining and Flow cytometry analysis. The biological function of QCR2 and PHB were determined using western blotting, RT-qPCR, microarray analysis and xenografts. The interactions between proteins and the ubiquitination of p53 were assessed by immunoprecipitation, mass spectrometry analysis and GST pull down. The subcellular location of PHB and QCR2 was assessed by immunoblotting and immunofluorescence. FINDING: The expression of QCR2 is upregulated in multiple human tumors. Suppression of QCR2 inhibits cancer cell growth by activating p53 signaling and inducing p21-dependent cell cycle arrest and senescence. QCR2 directly interacts with PHB in the mitochondria. Overexpression of QCR2 inhibits PHB binding to p53 in the nucleus, and facilitates p53 ubiquitination and degradation, consequently leading to tumorigenesis. Also, increased QCR2 and decreased PHB protein levels are well correlated with decreased expression of p21 in cervical cancer tissues. INTERPRETATION: These results identify a novel role for QCR2, together with PHB, in negative regulation of p53 stability and activity, thus promote cervical carcinogenesis. FUND: "973" Program of China, the National Science-technology Supporting Plan Projects, the National Natural Science Foundation of China, National Science and Technology Major Sub-Project and Technical Innovation Special Project of Hubei Province.

Loss of the novel mitochondrial protein FAM210B promotes metastasis via PDK4-dependent metabolic reprogramming.

Recent advances in tumor metabolism have revealed that metabolic reprogramming could dramatically promote caner metastasis. However, the relation and mechanism between metastasis and metabolic reprogramming are not thoroughly explored. Cell proliferation, colony formation, and invasion analysis were performed to evaluate the role of FAM210B in human cancer cells. Human ovarian cancer xenograft model was used to determine the effects of inhibiting FAM210B by shRNA on tumor metastasis. Microarray analysis was used to determine the target genes of FAM210B. FAM210B cellular localization was performed by mitochondria isolation and mitochondria protein extraction. To detect FAM210B-mediated metabolic reprogramming, oxygen consumption rate and extracellular acidification rate were measured. Our previous study screened a novel cancer progression-suppressor gene, FAM210B, which encodes an outer mitochondrial membrane protein, by the suppression of mortality by antisense rescue technique (SMART). Here we demonstrated that FAM210B loss was significantly associated with cancer metastasis and decreased survival in a clinical setting. Additionally, it was found that low expression of FAM210B was significantly correlated with decreased survival and enhanced metastasis in vivo and in vitro, and the loss of FAM210B led to an increased mitochondrial respiratory capacity and reduced glycolysis through the downregulation of pyruvate dehydrogenase kinase 4 (PDK4), which activated the EMT program and enhanced migratory and invasive properties. Collectively, our data unveil a potential metabolic target and mechanism of cancer metastasis.

Targeting CD146 in combination with vorinostat for the treatment of ovarian cancer cells.

Drug resistance is the predominant cause of mortality in late-stage patients with ovarian cancer. Histone deacetylase inhibitors (HDACis) have emerged as a novel type of second line drug with high specificity for tumor cells, including ovarian cancer cells. However, HDACis usually exhibit relatively low potencies when used as a single agent. The majority of current clinical trials are combination strategies. These strategies are more empirical than mechanism-based applications. Previously, it was reported that the adhesion molecule cluster of differentiation 146 (CD146) is significantly induced in HDACi-treated tumor cells. The present study additionally confirmed that the induction of CD146 is a common phenomenon in vorinostat-treated ovarian cancer cells. AA98, an anti-CD146 monoclonal antibody (mAb), was used to target CD146 function. Synergistic antitumoral effects between AA98 and vorinostat were examined in vitro and in vivo. The potential effect of combined AA98 and vorinostat treatment on the protein kinase B (Akt) pathway was determined by western blotting. The present study found that targeting of CD146 substantially enhanced vorinostat-induced killing via the suppression of activation of Akt pathways in ovarian cancer cells. AA98 in combination with vorinostat significantly inhibited cell proliferation and increased apoptosis. In vivo, AA98 synergized with vorinostat to inhibit tumor growth and prolong survival in ovarian cancer. These data suggest that an undesired induction of CD146 may serve as a protective response to offset the antitumor efficacy of vorinostat. By contrast, targeting CD146 in combination with vorinostat may be exploited as a novel strategy to more effectively kill ovarian cancer cells.

Geldanamycin, an inhibitor of Hsp90, increases paclitaxel-mediated toxicity in ovarian cancer cells through sustained activation of the p38/H2AX axis.

Paclitaxel is a mitotic inhibitor used in ovarian cancer chemotherapy. Unfortunately, due to the rapid genetic and epigenetic changes in adaptation to stress induced by anticancer drugs, cancer cells are often able to become resistant to single or multiple anticancer agents. However, it remains largely unknown how paclitaxel resistance happens. In this study, we generated a cell line of acquired resistance to paclitaxel therapy, A2780T, which is cross-resistant to other antimitotic drugs, such as PLK1 inhibitor or AURKA inhibitor. Immunoblotting revealed significant alterations in cell-cycle-related and apoptotic-related proteins involved in key signaling pathways. In particular, phosphorylation of p38, which activates H2AX, was significantly decreased in A2780T cells compared to the parental A2780 cells. Geldanamycin (GA), an inhibitor of Hsp90, sustained activation of the p38/H2AX axis, and A2780T cells were shown to be more sensitive to GA compared to A2780 cells. Furthermore, treatment of A2780 and A2780T cells with GA significantly enhanced sensitivity to paclitaxel. Meanwhile, GA cooperated with paclitaxel to suppress tumor growth in a mouse ovarian cancer xenograft model. In conclusion, GA may sensitize a subset of ovarian cancer to paclitaxel, particularly those tumors in which resistance is driven by inactivation of p38/H2AX axis.

The p38 MAPK inhibitor BIRB796 enhances the antitumor effects of VX680 in cervical cancer.

VX680 is a potent and selective inhibitor that targets the Aurora kinase family. The p38 mitogen-activated protein kinase (MAPK) regulates a large number of cellular pathways and plays an important role in the regulation of cell survival and apoptosis. This study aimed to evaluate the effect of VX680 on cervical cancer cells and investigate whether the effects on apoptosis are enhanced by the ablation of p38 MAPK activation. The results suggested that VX680 inhibited the proliferation of cervical cancer cells by causing G2/M phase arrest and endoreduplication and that the apoptotic effect was attenuated by the activation of p38 MAPK. However, the addition of BIRB796, which is an important p38 MAPK inhibitor, effectively eliminated the expression of p-p38 and hence significantly enhanced the cell death induced by VX680 in vitro. Further study demonstrated that BIRB796 cooperated with VX680 to suppress cervical cancer cell growth in a mouse xenograft model. Taken together, our results demonstrated that VX680 induced cell cycle arrest and endoreduplication in human cervical cancer cells. Combined treatment with VX680 and BIRB796 synergistically inhibited tumor growth both in vitro and in vivo. Dual blockade of Aurora kinases and p38 MAPK is therefore a promising strategy for cervical cancer treatment.

Effect of LIV1 on the sensitivity of ovarian cancer cells to trichostatin A.

In a previous study, we used a functional gene screen approach to identify the key genes responsible for the tumor-selective action of trichostatin A (TSA), of which LIV1, a novel zinc transporter, was isolated by its marked ability to confer resistance against TSA-induced apoptosis. The aim of the present study was to investigate the effect of LIV1 expression on the sensitivity of ovarian cancer cells to TSA. We tested the induction of LIV1 in ovarian cancer cells and clinical samples after TSA treatment by real-time PCR and western blot analysis. We investigated the effect of LIV1 expression on the sensitivity of ovarian cancer cells to TSA by MTT assay, flow cytometry and colony forming assays. Finally, we analyzed the mechanism of LIV1 in ovarian cancer cells by western blot analysis. We found that the levels of LIV1 mRNA and protein were significantly upregulated after TSA treatment. The viability and colony forming rates of the ovarian cancer cells transfected with AS-LIV1 (pCEP4 carrying antisense LIV1 cDNA) were obviously higher than the rates of the control as detected by MTT and colony forming assays, which could be reversed by FL-LIV1 (pCEP4 carrying full-length LIV1 cDNA). The apoptotic rate of the AS-LIV1 cells was markedly lower than the rate of the control as determined FACS. Using western blot analysis, we demostrated that the inhibition of TSA-induced apoptosis by knockdown of LIV1 might be associated with decreased endogenous levels of Bcl-2, enhanced levels of Bax and cleavage of procaspase-3. The present study suggests that the drug resistance of ovarian cancer cells to TSA may be related to expression of the LIV1 gene, and targeting LIV1 could be exploited as a novel strategy to more effectively kill ovarian cancer cells.

Ubiquitin B in cervical cancer: critical for the maintenance of cancer stem-like cell characters.

Cervical cancer cells exhibit an increased requirement for ubiquitin-dependent protein degradation associated with an elevated metabolic turnover rate. Ubiquitin, which is a small, highly conserved protein expressed in all eukaryotic cells, can be covalently linked to certain target proteins to mark them for degradation by the ubiquitin-proteasome system. Previous studies highlight the essential role of Ubiquitin B (UbB) and UbB-dependent proteasomal protein degradation in histone deacetylase inhibitor (HDACi) -induced tumor selectivity. We hypothesized that UbB plays a critical role in the function of cervical cancer stem cells. We measured endogenous UbB levels in mammospheres in vitro by real-time PCR and Western blotting. The function of UbB in cancer stem-like cells was assessed after knockdown of UbB expression in prolonged Trichostatin A-selected HeLa cells (HeLa/TSA) by measuring in vitro cell proliferation, cell apoptosis, invasion, and chemotherapy resistance as well as by measuring in vivo growth in an orthotopic model of cervical cancer. We also assessed the cancer stem cell frequency, tumorsphere formation, and in vivo growth of human cervical cancer xenografts after UbB silencing. We found that HeLa/TSA were resistant to chemotherapy, highly expressed the UbB gene and the stem cell markers Sox2, Oct4 and Nanog. These cells also displayed induced differentiation abilities, including enhanced migration/invasion/malignancy capabilities in vitro and in vivo. Furthermore, an elevated expression of UbB was shown in the tumor samples of chemotherapy patients. Silencing of UbB inhibited tumorsphere formation, lowered the expression of stem cell markers and decreased cervical xenograft growth. Our results demonstrate that UbB was significantly increased in prolonged Trichostatin A-selected HeLa cells and it played a key role in the maintenance of cervical cancer stem-like cells.

Inhibition of Hec1 expression enhances the sensitivity of human ovarian cancer cells to paclitaxel.

AIM: Hec1, a member of the Ndc80 kinetochore complex, is highly expressed in cancers. The aim of this study was to explore the role and mechanism of action of Hec1 with respect to the cytotoxicity of paclitaxel in ovarian cancer. METHODS: Thirty ovarian cancer samples and 6 normal ovarian samples were collected. Hec1 expression in these samples was determined with immunohistochemistry. Ovarian cancer cell lines A2780, OV2008, C13K, SKOV3, and CAOV3 and A2780/Taxol were examined. Cell apoptosis and cell cycle analysis were detected with flow cytometric technique. siRNA was used to delete Hec1 in the cells. The expression of related mRNAs and proteins was measured using RT-PCR and Western blot analysis, respectively. RESULTS: Hec1 expression was significantly higher in ovarian cancer samples than in normal ovarian samples, and was associated with paclitaxel-resistance and poor prognosis. Among the 6 ovarian cancer cell lines examined, Hec1 expression was highest in paclitaxel-resistant A2780/Taxol cells, and lowest in A2780 cells. Depleting Hec1 in A2780/Taxol cells with siRNA decreased the IC50 value of paclitaxel by more than 10-fold (from 590±26.7 to 45.6±19.4 nmol/L). Depleting Hec1 in A2780 cells had no significant effect on the paclitaxel sensitivity. In paclitaxel-treated A2780/Taxol cells, depleting Hec1 significantly increased the cleaved PARP and Bax protein levels, and decreased the Bcl-xL protein level. CONCLUSION: Hec1 overexpression is associated with the progression and poor prognosis of ovarian cancer. Inhibition of Hec1 expression can sensitize ovarian cancer cells to paclitaxel.

Breast cancer associated fibroblasts promote MCF-7 invasion in vitro by secretion of HGF.

This study was aimed to explore the influence of breast cancer associated fibroblasts (CAFs) in migration and invasion of breast cancer cell line MCF-7, and investigate whether hepatocyte growth factor (HGF) is involved in this process. Primary breast CAFs and their corresponding normal breast fibroblasts (NFs) were obtained by collagenase digestion. On the basis of the co-culture, the migration and invasion capacity of MCF-7 cells was compared between CAFs and NFs by Transwell. The difference in the HGF expression between them was detected by ELISA. The secretion of HGF was knocked down by using RNA interference technology in CAFs. Then the changes of migration and invasion capacity of MCF-7 cells were investigated by Transwell. Eventually, we isolated high-purity CAFs and NFs, and the CAFs had a stronger ability in promoting MCF-7 migration and invasion than the NFs. ELISA results demonstrated that CAFs secreted higher HGF, and the capacity of MCF-7 migration and invasion was declined after knocking down the secretion of HGF in CAFs by RNA interference. It is suggested that CAFs can promote MCF-7 migration and invasion through HGF in vitro.