Page 151 - SRPSKO DRUŠTVO ISTRAŽIVAČA RAKA
P. 151

SDIRSACR                                                                                 Oncology Insights


        P59

        The critical role of ADAMTS8 expression deregulation in forwarding vascular invasion in rectal carcinoma

        Bojana Kožik , Tarik Čorbo , Naris Pojskić , Lejla Kapur Pojskić , Lidija Todorović , Ana Božović , Ana Kolaković 1
                                                                              1
                   1
                               2
                                                                                          1
                                                              2
                                            2
        1Laboratory for Radiobiology and Molecular Genetics, Vinča Institute of Nuclear Sciences, National Institute of Republic of Serbia,
        University of Belgrade, Belgrade, Serbia
        2Institute for Genetic Engineering and Biotechnology, University of Sarajevo, Sarajevo, Bosnia and Herzegovina
        Keywords: rectal carcinoma, vascular invasion, ADAMTS8

        Background: Rectal carcinoma (RC) accounts for approximately 30% of all colorectal carcinomas (CRC) and is recognized
        as a distinct clinical entity with unique causes and treatment strategies. Aside from lymphatic invasion, RC can spread in
        several vascular ways previously identified and validated as prognostic markers. Vascular invasion (VI) has consistently
        been shown to be associated with poor prognosis, both when detected by pathology and radiology methods, however,
        the molecular basis of VI is poorly studied. Several mechanisms may contribute to the VI process, including epithelial-
        mesenchymal  transition  (EMT),  remodeling  of  the  extracellular  matrix  (ECM),  and  angiogenesis.  This  integrative
        study was conducted with the aim of identifying genes and pathways specific for a VI-positive rectal adenocarcinoma
        phenotype.
        Material and Methods: Using cBioPortal dataset platform, we examined mRNA expression and methylation data of 41
        pre-selected, literature-based genes proven to be involved in regulating EMT, ECM, and angiogenesis during colorectal
        carcinogenesis and their potential association with vascular invasion as a clinical feature in a RC dataset consisting of
        78 samples, which were filtered from the COADREAD TCGA (Firehose Legacy) dataset.
        Results: Among 41 tested gene candidates, only ADAMTS8 showed a significant association between lower ADAMTS8
        mRNA expression and vascular invasion was detected (p=0.039, tested by Mann-Whitney U test). In addition, methylome
        analysis strongly suggests promoter methylation of the ADAMTS8 gene, as a mechanism of epigenetic gene silencing,
        since in this cohort ADAMTS8 mRNA expression significantly correlates with ADAMTS8 methylation level (p=0.012, r=-
        0.263, Spearman’s test of correlation). Moreover, we observed a tendency toward association between higher level of
        ADAMTS8 methylation and vascular invasion (p=0.087, tested by Mann-Whitney U test). Interestingly, protein-protein
        interaction analysis among protein products of tested 41 genes on STRING platform revealed that only ADAMTS8
        protein formed a distinct cluster, displaying no interactions with any other queried proteins. Observed associations
        require further functional and experimental analysis to elucidate the exact role of ADAMTS8 expression deregulation
        in forwarding vascular invasion RC.
        Conclusions: Obtained results have potential clinical utility, providing new approaches for developing targeted therapies
        for VI-positive rectal cancer, and require in vitro validation.

        Acknowledgments and funding This work is the result of collaboration within COST Translacore CA21154.





        P60

        Deep learning on combined multistain and color-depth histology enhances early breast cancer prognosis

        Yifei Lin1, Xingyu Li1, Jelena Milovanovic2, Nataša Todorovic Rakovic2, Velicko Vranes3, Tijana Vujasinović2, Ksenija
        Kanjer2, Marko Radulovic2

        1Electrical & Computer Engineering Department, Faculty of Engineering, University of Alberta, Edmonton, Canada
        2Department of Experimental Oncology, Institute of Oncology and Radiology of Serbia, Belgrade, Serbia
        3Department of Basic and Environmental Science, Instituto Tecnológico de Santo Domingo (INTEC), Santo Domingo, Dominican
        Republic

        Keywords: breast cancer, deep learning, distant metastasis, histopathology, microscopy, multicolor-depth, multistain,
        pan-cytokeratin, prognosis, ResNet

        Background: Traditional clinicopathological markers of breast cancer outcome are now supplemented by molecular
        assays such as gene-expression panels, microRNA signatures, circulating tumour cells, proliferation markers, cytokines,


   136
   146   147   148   149   150   151   152   153   154   155   156