Search

Glioblastoma is the most aggressive type of brain cancer with high-levels of intra- and inter-tumour heterogeneity that contribute to its rapid growth and invasion within the brain. Here, we have used a deep convolutional neural network (DCNN) as a semantic segmentation model to segment seven different tumour regions including leading-edge (LE), infiltrating tumour (IT), cellular tumour (CT), cellular tumour microvascular proliferation (CTmvp), cellular tumour pseudopalisading region around necrosis (CTpan), cellular tumour perinecrotic zones (CTpnz) and cellular tumour necrosis (CTne) in digitised glioblastoma histopathological slides from The Cancer Genome Atlas (TCGA). Analysis of segmentation results from tumour images together with matched RNA expression data identified genetic signatures that are specific to these different tumour regions. We found that spatially resolved gene signatures were strongly correlated with survival in patients with defined genetic mutations. Moreover, in silico cell ontology analysis and single-cell RNA sequencing data from resected glioblastoma tissue samples, showed that these tumour regions had different gene signatures, suggesting they are driven by different cell types in the tumour microenvironment. This points to a key role for interactions between microglia/pericytes/monocytes and tumour cells that occur in the IT and CTmvp regions, which may contribute to poor patient survival. Overall, this work identifies key histopathological features that are indicative of patient survival and detected spatially associated genetic signatures that mediate tumour-stroma interactions that should be investigated as new targets in glioblastoma. Citation Format: Amin Zadeh Shirazi, Mark D. McDonnell, Eric Fornaciari, Narjes Sadat Bagherian, Kaitlin G. Scheer, Michael S. Samuel, Mahdi Yaghoobi, Rebecca J. Ormsby, Santosh Poonnoose, Damon Tumes, Guillermo A. Gomez. A deep convolutional neural network for segmentation of whole-slide pathology images in glioblastoma [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-004.

Pancreatic cancer is one of the leading causes of cancer death and despite advances in chemotherapeutic regimens the overall 5-year survival rate remains less than 5%. The actomyosin-regulating ROCK1 and ROCK2 kinases are downstream-targets of the Rho GTPase pathway. They contribute to various processes, such as cell adhesion, motility, proliferation, differentiation and survival, which then influence numerous stages of cancer growth and progression. Interestingly, exome sequencing of pancreatic cancer genomes revealed that 15% of pancreatic cancer patients carry an amplification of the ROCK1 gene (1). Moreover, we found significant increases in ROCK1 and ROCK2 RNA expression in pancreatic cancer datasets obtained using Oncomine. When we further determined the protein expression of ROCK2 in pancreatic ductal adenocarcinoma (PDAC) we found an up-regulation of ROCK2 during disease progression. Higher ROCK2 levels also correlated with less differentiated tumors. A characteristic of advanced stages of pancreatic cancer is a collagen and fibroblast enriched stroma. To analyze the effect of ROCK kinases on cancer cell invasion, we performed 3-dimensional organotypic assays. Our results demonstrate that increased ROCK signaling was sufficient to convert non-invasive PDAC cells into ones capable of invasion into organotypic collagen matrix. Using an RNA sequencing approach, we investigated the effect of ROCK activation on pancreatic cancer cell gene transcription. Differential expression analysis revealed an enrichment of gene sets that are involved in cell-matrix interaction, and we found a higher release of matrix metalloproteinases MMP10 and MMP13 upon ROCK activation. In addition, our organotypic studies revealed extensive tissue remodeling and an accumulation of cleaved collagen bundles at the sites of PDAC cell invasion. Furthermore, collagen degradation and cell invasion into organotypic matrices were significantly reduced by application of a broad-spectrum MMP inhibitor, confirming that ROCK-induced invasion is dependent on MMP activity. Interestingly, our studies also indicated that ROCK-driven invasion of PDAC cells into collagen matrix enabled cell growth. To study the function of ROCK kinase signaling in pancreatic cancer in vivo, we expressed conditionally active ROCK2 in a mouse model of PDAC. We found that an increase of ROCK activity in pancreatic cancer cells accelerated PDAC progression, which resulted in reduced survival. Contrary, the administration of a ROCK inhibitor during tumor progression had a beneficial effect on survival. In summary, our results suggest that targeting ROCK kinases should be considered for chemotherapy of invasive pancreatic cancer. (1) Biankin et al. Nature. 2012 Nov 15;491(7424):399-405 Citation Format: Nicola Rath, Shereen Kadir, Jennifer P. Morton, Andreia V. Pinho, Lena Helbig, Linda Julian, Ewan J. McGhee, Gabriela Kalna, Alexei Vazquez, Kurt I. Anderson, Ilse Rooman, Michael S. Samuel, Michael F. Olson. ROCK kinases drive invasive pancreatic tumor growth. [abstract]. In: Proceedings of the AACR Special Conference: Function of Tumor Microenvironment in Cancer Progression; 2016 Jan 7–10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2016;76(15 Suppl):Abstract nr C22.

To study tumour progression mechanism in vivo, ROCK 2 signalling deregulation and co-operation with activated rasHa or PTEN loss was investigated in transgenic mouse skin carcinogenesis. Transgenic mice that expressed a 4-hydroxytamoxifen (4HT)-activated human ROCK 2-estrogen receptor fusion transgene from a keratin 14 promoter [K14.ROCKer] were crossed to mice expressing activated rasHa exclusively in epidermal transit amplifying keratinocytes [HK1.ras-ROCKer] or mice where cutaneous PTEN loss was achieved in basal layers, hair follicles and stem cell keratinocytes via topical RU486 treatment of bi-genic K14.creP/Δ5PTENflx genotypes [K14.Δ5PTEN-ROCKer]. Initial 4HT-treatments of K14.ROCKer mice [3/wk; 26 wks] induced epidermal and follicular hyperplasia but no papillomas; whilst treatment of HK1.ras line 1205 gave ear tag papillomas [10-12 wks] typically smaller than vehicle controls, but no malignant conversion. In contrast, 4-HT treated HK1.ras-Rocker papillomas [10-12 wks] exhibited areas of carcinoma in situ and well-differentiated squamous cell carcinoma [wdSCC]. With time [16-20wks] wdSCC areas increased, concomitant with loss of p53, increased p-AKT and altered differentiation marker expression; however histotypes remained wdSCC despite continued 4HT-induced Rock 2 activity; as confirmed by downstream MYPT-1 phosphorylation. Furthermore, papillomas regressed whenever 4-HT treated HK1.ras-Rocker mice lost ear tags, suggesting that continued promotion from wounding during papillomatogenesis was critical to achieving the appropriate [late-stage] papilloma context that facilitated Rock 2-mediated conversion. To test this idea, K14.ROCKer mice were crossed to HK1.ras line 1276, which is insensitive to wound promotion, and such bi-genic mice did not exhibit papillomas; while in reverse, demonstrating the need for initiation, classic TPA-promoted K14.ROCKer mice were also devoid of papillomas. In addition, as previously rasHa/PTENnull mice exhibited papillomas prone to conversion on acquisition of an additional oncogene [e.g. fos], K14.ROCKer mice were bred into a K14.cre.Δ5PTEN strain and bi-genic mice treated with Ru486 [3wks] and 4HT [>26wks]. Again, in the absence of rasHa no papillomas were observed and mice exhibited increased hyperplasia and altered differentiation consistent with disruption of epidermal physiology. These data suggest that Rock 2 deregulation plays major roles in malignant conversion and progression, and in cooperation with rasHa activation or PTEN loss, the mechanism requires additional promotion/initiation events during papillomatogenesis to create the context of late-stage papilloma where Rock 2 activities become causal; an idea being tested in tri-genic HK1.ras/Δ5PTEN-Rocker genotypes. Citation Format: Siti F. Masre, Michael S. Samuel, Michael F. Olson, David A. Greenhalgh. Inducible ROCK 2/rasHa cooperation requires wound promotion to achieve malignancy in transgenic mouse skin carcinogenesis, whereas inducible ROCK 2/PTEN loss fails to achieve benign papilloma. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 930. doi:10.1158/1538-7445.AM2014-930