Abstracts of the posters to be presented at the symposium on May 8, 2019, with poster number, name of the presenter and the poster title.
HERCULES project partners:
1) University of Helsinki, Finland: Sampsa Hautaniemi (coordinator); Tero Aittokallio; Krister Wennerberg; Anna Vähärautio. 2) Hospital District of Southwest Finland: Seija Grénman. 3) University of Turku, Finland: Olli Carpén. 4) University of Trieste, Italy: Serena Bonin. 5) Istituto Superiore di Sanita, Italy: Mauro Biﬀoni.
6) University of Cambridge, UK: Jussi Taipale. 7) Institute Pasteur, France: Benno Schwikowski. 8) AB Analitica SRL, Italy: Dino Paladin.
Background: High-grade serous ovarian cancer (HGSOC) is the most diﬃcult to treat subtype of all ovarian cancers. Genetic heterogeneity makes ﬁnding of drugs that are able to kill all the cancer cell populations challenging because some of them may be resistant already at diagnosis or become resistant during treatment.
Project overview and aims: The massive amounts of sequencing and other molecular data produced in HERCULES together with extensive clinical information enable identification of novel biomarkers and features of drug resistant cell populations to suggest effective combinatorial therapies. Designated computational models and tools to analyze the massive amount of produced data are developed in the project.
Status May 2019: Altogether 1089 samples from 216 ovarian cancer patients have been stored so far. High-throughput molecular data produced from over 60 patients include over 600 samples i.a. DNA (n=270) and RNA (n=238) sequencing, single-cell RNA sequencing (n=70), mass cytometry (n=35), and ex vivo drug testing (n=34).
Examples of new views to data: In order to improve identification of genetic and molecular features affecting chemotherapy response, we have applied novel criteria to group patients to the poorest and best responders. Additionally, global differences between “early” (stage I-II sites) and “late” (stage III-IV) tumors, and evolutionary trees (phylogenetics) are now used to improve understanding of genomics data.
Funding: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 667403.
Tumors of most high-grade serous ovarian cancer patients display either initial or acquired therapy resistance. The CHEMORESPONSE consortium, consisting of the groups of Sampsa Hautaniemi, Sakari Hietanen, Liisa Kauppi (consortium leader), Mikko Niemi and Eija Pirinen, addresses the urgent need for patient-specific prospective prediction of the most effective cancer therapy regimen. Anti-cancer drugs induce a multitude of cellular changes that alter DNA repair pathways, metabolism and membrane transporter function. The CHEMOREPONSE consortium aims to dissect the interplay between them by performing cutting-edge molecular measurements from patient cells. We focus on the effects of platinum agents and PARP inhibitors. Based on the molecular read-outs, combined with comprehensive clinical and genomic data, we will develop a predictive model for patient-specific responses to anti-cancer drugs. This will set the stage for tailored ovarian cancer therapy with maximal efficacy and minimal toxicity.
Funding: This project has received funding from the Academy of Finland and the Cancer Foundation Finland under the Health from Science (TERVA) program.
Authors: Anniina Farkkila1,2, Julia Casado3, Doga Gulhan2, Connor Jacobson4, Huy Nguyen1, Bose Koruchupakkal1, Zoltan Maliga4, Jia R. Lin4, Yinghui Zhou5, Julie R. Graham5, Bruce J. Dezube5, Steven Waggoner6, Pamela Munster7, Gini F. Fleming8, Sandro Santagata9, Ursula A. Matulonis1, Peter Park1, Sampsa Hautaniemi3, Peter K. Sorger4, Elizabeth M. Swisher10, Alan D. D’Andrea1, and Panagiotis Konstantinopoulos1.
Author affiliations: 1Dana-Farber Cancer Institute, Boston, MA; 2Harvard Medical School, Boston, MA; 3 Oncosys research program, University of Helsinki, Helsinki, Finland, 4Laboratory of Systems Pharmacology, Harvard Medical School, Boston, MA, 5Tesaro, Waltham, MA; 6University Hospitals of Cleveland, Cleveland, OH; 7Helen Diller Family Comprehensive Cancer Center, San Francisco, CA; 8University of Chicago, Chicago, IL; 9Brigham and Women's Hospital, Laboratory for Systems Pharmacology, Boston, MA; 10University of Washington, Seattle, WA
Combining Poly-ADP Ribose Polymerase (PARP) inhibitors to immune checkpoint blockade (ICB) has emerged as a promising approach in ovarian cancer (OC). TOPACIO/Keynote-162 is a multi-center, open-label, single-arm phase 1/2 trial of Poly-ADP Ribose Polymerase (PARP) inhibitor niraparib combined with an anti-PD-1 antibody pembrolizumab that included 62 patients with relapsed, platinum resistant OC. Overall, the confirmed objective response (OR) rate was 18% with a clinical benefit (complete response + partial response + stable disease; CB) rate of 65%. We analyzed pre-trial Formalin Fixed Paraffin Embedded tumor samples for genomic and immunologic determinants of response. We found that tumor mutational signature 3 significantly associated with CB, and correlated with increased progression-free survival. In immuneprofiling with Nanostring gene expression analysis, increased scores for Type-I interferon pathway activation, and exhausted CD8+T-cell signature significantly associated with OR. Comprehensive analysis of the cell types, functional states, and spatial interactions using a novel highly multiplexed tissue immunofluorescence (tCycIF) revealed additional determinants of response. In conclusion, combination of Niraparib and Pembrolizumab shows promising clinical activity in relapsed platinum resistant OC, and patient stratification using combined immunogenomic biomarkers can potentially further increase the efficacy.
Authors: Wojciech Senkowski1, Daria Bulanova2, Krister Wennerberg1,2
Author affiliations: 1Biotech Research & Innovation Centre, University of Copenhagen, Denmark; 2Institute for Molecular Medicine Finland, University of Helsinki, Finland
Organoids have been proposed for application in functional precision medicine, as they mimic heterogeneity and drug response of in vivo tumours more closely than standard cell cultures. However, ovarian cancer organoids (OvCaOs) have been poorly described so far. Thus, we set out to establish a new OvCaO culture method, that would allow the expansion of cellular material and maintenance of long-term growth.
Here, we evaluated the growth and PAX8 expression of multiple cryopreserved ovarian cancer samples under a number of different organoid media conditions. Our experiments revealed that some growth factors and media supplements commonly used in organoid culture have detrimental effects on the growth of OvCaOs. For instance, the addition of EGF or FGF-2 often caused the cells to differentiate, which correlated with the loss of PAX8 nuclear staining. On the contrary, we found some media supplements, such as SB202190 (inhibitor of p38) or FGF-10 to promote organoid growth. Interestingly, we also found that growth factors that were not previously reported in organoid culture, such as FGF-4, IGF-I or VEGF, demonstrate beneficial effects on OvCaO growth.
In summary, our work reveals previously unrecognized effects of various growth factors in the OvCaO culture and demonstrates the necessity of careful consideration of organoid culture conditions. Currently, we continue the experiments on designing the optimal ovarian cancer organoid medium for long-term culture of OvCaOs.
Authors: Sini Pirnes-Karhu 1, Virpi Asikainen 1, Minna Tuominen 1, Manuela Tumiati 2, Johanna Hynninen 3, Sakari Hietanen 3, Liisa Kauppi 2, Eija Pirinen 1
Author affiliations: 1 Research Program for Clinical and Molecular Medicine, Faculty of Medicine, University of Helsinki, Finland; 2 Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Finland; 3Department of Obstetrics and Gynaecology, Turku University Hospital, Finland.
In recent years, studies have revealed the capacity of cancer cells for both glycolysis and mitochondrial oxidative phosphorylation (OXPHOS) and metabolic heterogeneity even within tumors of same clinical diagnosis is apparent. To elucidate metabolic differences between high-grade serous ovarian cancer (HGSOC) patients, we have developed an ex vivo analysis to measure mitochondrial OXPHOS from fresh HGSOC tumors and healthy control samples by using high-resolution respirometry. Different tissue processing and respiratory analysis conditions were tested to suit HGSOC and control samples of varying consistency from different anatomical sites. Mechanical permeabilization with forceps and scissors proved to be most suitable as it was gentle enough to preserve mitochondrial membrane integrity but allowed respiratory substrates to reach OXPHOS complexes. OXPHOS was measured successfully from small tissue pieces of five to seventeen milligrams. Our preliminary results show that the OXPHOS profile was different between HGSOC tumor and control samples, mitochondrial complex IV-linked respiration being higher in tumors.
Altogether, we have established a reliable method to measure OXPHOS from freshly collected HGSOC and control tissue with heterogenous consistency. This method reflects in vivo mitochondrial function of the tissues, as the samples are minimally processed, and serves as a valuable tool to study mitochondrial metabolism in ovarian cancer.
Authors: Virpi Asikainen1, Sini Pirnes-Karhu1, Minna Tuominen1, Johanna Hynninen2, Sakari Hietanen2, Liisa Kauppi1, Eija Pirinen1
Author affiliations: 1University of Helsinki,Finland, 2Turku University Hospital, Finland.
Cancer development and progression have been associated with changes in cellular metabolism including abnormalities in mitochondrial function and mitochondrial DNA (mtDNA) amount. In ovarian cancer, mtDNA amount is shown to vary depending on the type and grade of the cancer, which can affect the mitochondrial oxidative capacity and the drug response. Our aim is to understand the impact of mtDNA amount on cellular metabolism and cancer drug efficacy. Here we focused on developing quantitative PCR (qPCR) method to assess mtDNA amount from ovarian cancer and healthy control samples. With qPCR three mitochondrial genes were amplified and results were normalized using genomic DNA amount. Different homogenization methods were tested to fit all samples with varying consistency. From a DNA yield and integrity perspective, homogenization with TissueLyser was observed to be the most optimal choice. Based on cycle quantification (Cq) values obtained with qPCR, we were able to measure mtDNA amount successfully from tumor and healthy control tissue pieces as small as five milligrams within desired Cq value range. Our preliminary results showed variation in mtDNA amount between different cancer types. Overall, we have established a method to study mtDNA amount in ovarian cancer samples. Together with mitochondrial functional measurement, mtDNA amount assessment can offer valuable information about cellular metabolism in ovarian cancer samples.
Authors: Eros Azzalini1,2, Anna Sapino3,4, Caterina Marchiò3,4, Umberto Miglio4, Vincenzo Canzonieri1,5, Giorgio Stanta1 and Serena Bonin1
Author affiliations: 1: Department of Medical Sciences-University of Trieste- Cattinara Hospital- Trieste, Italy. 2: Doctorate of Nanotechnology, University of Trieste, Trieste, Italy. 3: Department of Medical Sciences, University of Torino, Torino, Italy. 4: Candiolo Cancer Institute-FPO, IRCCS, Candiolo, Italy. 5: CRO- Centro di Riferimento Oncologico- Aviano, Italy.
BACKGROUND: Tissue blocks fixed in Bouin’s solution (BS) are stored in several European repositories and hospitals representing a valuable resource for biomarkers validation. We assessed the suitability of Bouin’s fixed tissues for RNA expression analyses in light of the recent cutting-edge technologies.
METHODS: Bouin’s and formalin fixed specimens were compared in 15 matched cases of high grade serous ovarian cancer (HGSC) debulking surgeries. Total RNA yield and quality were measured by Nanodrop, RNA integrity was assessed by Agilent BioAnalyzer and mRNA expression was investigated by qRT-PCR, Nanostring and ddPCR. The resulting data were then correlated among the three platforms.
RESULTS: No significant difference was found between the two fixatives regarding RNA yield, but RNA from Bouin’s samples showed a higher level of fragmentation and lower quality when compared to formalin. The mean expression levels of the genes studied were higher in formalin specimens in comparison to BS in all the systems analyzed.
Authors: Diego Boscarino, Isabella Caligiuri, Mauro Simonato.
Author affiliations: AB ANALITICA SRL, Italy
AB ANALITICA SRL is a company specialized in IVD (In Vitro Diagnostic) devices based on molecular diagnostics technologies and compliant with the requirements of EU regulations applicable to such devices (CE marking certifying conformity to the IVD Directive 98/79/EC). AB ANALITICA SRL is ISO 13485 certified, the standard for quality management systems applied by companies operating in this market, addressing all company processes including design and development processes. The development of a clinically applicable tissue-based test in the HERCULES project has been conducted in the framework of such requirements.
The determination of product requirements included a patent analysis and a feasibility study addressing the definition of technical specifications and final identification of NGS (Next Generation Sequencing) as the assay technology, requirements for regulatory compliance, a market search and competitors analysis, and an assessment of technical and financial viability.
Since validation of the whole workflow including pre-examination process is fundamental to accomplish safety and effectiveness of the IVD device, test development started with a study of optimal DNA/RNA extraction conditions, tied to NGS nucleic acid input requirements, in which after an initial screening of more than 40 extraction protocols, 12 were extensively investigated and ranked.
The panel of target genes has been updated based on recent advances in ovarian cancer management, and NGS assay development is in progress.
Authors: Valeria Ariotta1, Jimmy Azar1, Elisa Ficarra2, Annika Auranen3, Olli Carpén1,4,5, Sampsa Hautaniemi1
Author affiliations: 1Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Helsinki, Finland. 2Dept. of Computer and Control Engineering, Politecnico di Torino, Torino, Italy. 3Dept. of Gynaecology and Obstetrics, University of Tampere, Finland. 4 Department of Pathology, University of Turku and Turku University Hospital, Turku, Finland. 5Auria Biobank, University of Turku and Turku University Hospital, Turku, Finland.
We have developed an open-source computational framework to analyze and predict treatment response based on digitalized histopathological images. The samples are treated with hematoxylin and eosin stain (H&E), the former coloring the cell nuclei with blue/purple shades and the latter coloring the cytoplasm and the connective tissue in pink/red shades. The utility of the framework is shown to predict chemotherapy response of 37 high-grade serous ovarian cancer (HGSOC) patients from HERCULES cohort.
The standard treatment for HGSOC consists of surgery and platinum-taxane chemotherapy and even though 80% of patients have an excellent initial response, the majority relapse within 18 months leading to less than 45% 5-year survival rate. Thus, it is important to develop tools to predict the patient’s response and identify which patients will benefit from the treatment.
The study is mainly composed by two phases. Firstly, tools have been developed to extract features, such as Haralick and local binary patterns (LBP) features, by means of texture methods, and morphological feature related to the density and the heterogeneity of the cells, starting from H&E images. Secondly, we used ensemble decision tree algorithm to predict response to chemotherapy based on the features extracted from the digitalized H&E stained images and achieved prediction accuracy of 83%.
Authors: Riikka J Niemi1, Elena I Braicu2, Hagen Kulbe2, Kaisa M Koistinen3, Jalid Sehouli2, Ulla Puistola4, Johanna U Mäenpää4, Mika Hilvo3
Author affiliations: 1Department of Obstetrics and Gynecology, Tampere University Hospital, Tampere, Finland; 2Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin Humboldt-Universität zu Berlin, and Berlin Institute of Health, Department of Gynecology, Berlin, Germany; 3Zora Biosciences Oy, Espoo, Finland; 4Department of Obstetrics and Gynecology, PEDEGO Research Unit, Medical Research Center Oulu, University of Oulu and University Hospital of Oulu, Oulu, Finland; 5Faculty of Medicine and Life Sciences, University of Tampere, Tampere, Finland.
Background: Different histological subtypes of ovarian cancer have distinct mutational spectrum, but it is not known whether the subtypes share common metabolic abnormalities. Previously, we have shown large lipid metabolism alterations in the serum of late-stage, high-grade serous ovarian carcinoma (HGSOC) patients. To validate these findings and investigate the metabolic alterations further, we investigated lipidomic changes in patients with tumors of early-stage and various histological subtype.
Methods: Altogether, 354 serum/plasma samples were collected from three centers, one from Germany and two from Finland. We performed lipidomic analysis of samples from patients with malignant (N=138) or borderline (N=25) ovarian tumors, and 191 controls with benign pathology, and compared the results to previously published data
Results: We found large number of lipids, including e.g. phospholipids, ceramides and triacylglycerols, which showed consistent alteration both in early- and late-stage ovarian cancer patients as well as in pre- and postmenopausal women. Most of these changes were already significant at an early stage and progressed with increasing stage. Furthermore, many lipids showed similar alterations in serous, mucinous and endometrioid histological subtypes.
Conclusions: Changes in lipid metabolism occur also in other histological subtypes besides HGSOC, and can be observed already in early-stage disease. This may enable new diagnostic and therapeutic opportunities.
Authors: Jani Salmi, Anna-Maria Mende, Taru Tuomi, Mikko Loukovaara, Olli Carpén
Author affiliations: Research Program in Systems Oncology, University of Helsinki. Helsinki Biobank, HUSLAB, Helsinki University Hospital. Department of Obstetrics and Gynecology, University of Helsinki and Helsinki University Hospital
The overall 5-year survival in high-grade serous ovarian cancer (HGSOC) is below 40%, however, significant variation from primary chemoresistance to extended survival and cure is seen. Electronic medical record-based information could be used to identify features that predict response to first-line chemotherapy and/or survival. We collected information from 924 HGSOC patients treated at Helsinki University Hospital during years 1987-2013 and used computational tools to generate disease trajectories and create models for outcomes. In stage III-IV patients (85% of the cohort), primary treatment strategy and surgery result strongly associated with overall survival. For patients with optimal primary debulking surgery result (59% of the cohort), 5-y overall survival rate was 81%, whereas only 22% of patients with interval debulking and suboptimal result (14% of the cohort) were alive at 5 years. Importantly, extended outcomes were seen in all groups independent of treatment strategy or disease trajectory. Two patterns for the extended survival were observed. For some patients the disease did not relapse after primary treatment, whereas for others multiple relapses and remissions were seen. These subgroups can be further defined with additional clinical data. These results highlight the need for detailed clinical information and patient stratification, when studying the biological features associated with HGSOC behavior and in search for prognostic biomarkers.
Authors: Isoviita Veli-Matti1, Salminen Liina2, Azar Jimmy1, Lehtonen Rainer1, Roering Pia3, Carpén Olli4, Hietanen Sakari2, Grénman Seija2, Hynninen Johanna2, Färkkilä Anniina1,5, Hautaniemi Sampsa1
Author affiliations: 1Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Finland, 2Department of Obstetrics and Gynaecology, University of Turku and Turku University Hospital, Finland, 3Department of Pathology and Forensic Medicine, University of Turku, Finland, 4Institute of Biomedicine, Research Center for Cancer, Infections and Immunity, University of Turku, Finland, 5Department of Radiation Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA.
We have created a cloud-based machine learning system (CLOBNET) that is an open source, lean infrastructure for electronic health record (EHR) data integration capable of extract, transform, and load (ETL) processing. CLOBNET enables comprehensive analysis and visualization of structured EHR data. We demonstrate the utility of CLOBNET by predicting primary therapy outcomes of high-grade serous ovarian cancer (HGSOC) patients based on EHR data.
Materials and Methods
CLOBNET is built using open-source software to make data preprocessing, analysis, and model training user friendly. The source code of CLOBNET is available in GitHub. The HGSOC dataset was based on a prospective cohort of 208 HGSOC patients treated at Turku University Hospital, Finland from 2009 to 2019 for whom comprehensive clinical and EHR data were available.
We trained machine learning (ML) models using clinical data including a herein developed dissemination score that quantifies the disease burden at the time of diagnosis to identify patients with progressive disease (PD) or a complete response (CR) based on Response Evaluation Criteria in Solid Tumors (RECIST 1.1). The best performance was achieved with a logistic regression model, which resulted in an AUROC of 0.86 with specificity of 73% and sensitivity of 89%, when classifying between PD and CR patients. Our results demonstrate that CLOBNET allows predictions to be made based on EHR data to address clinically relevant questions.
Authors: Joonas Jukonen1, Katrin Höpfner1, Elina Pietilä1, Lidia Moyano Galceran2, Ping Chen1, Laura Lehtinen3, Kaiyang Zhang1, Kaisa Huhtinen3, Sampsa Hautaniemi1, Olli Carpén4, Kaisa Lehti1,2
Author affiliations:1: Research Programs Unit, University of Helsinki, Finland. 2: Department of Microbiology, Tumor and Cell Biology, Karolinska Institutet, Sweden. 3: Department of Pathology, University of Turku and Turku University Hospital, Finland. 4: Pathology, Research Programs Unit and HUSLAB, University of Helsinki and Helsinki University Hospital, Finland; Auria Biobank, University of Turku and Turku University Hospital, Finland.
Receptor tyrosine kinase signaling is implicated in multiple cancers, but various receptor tyrosine kinase families are poorly characterized in ovarian cancer. Our study provides experimental and clinical data regarding oncogenic receptor tyrosine kinase signaling in high-grade serous ovarian cancer.
Survival association data for candidate genes was obtained by analyzing a high-grade serous ovarian cancer mRNA dataset from The Cancer Genome Atlas. Functions of genes of interest were investigated using ovarian cancer cell lines. Genes of interest were studied in a clinical setting using a tumor microarray and a primary longitudinal sample set.
Several significant survival associations were discovered. Functional studies revealed novel data regarding genes of interest. Analysis of an epithelial ovarian cancer microarray revealed the association of a gene of interest in the high-grade serous subtype, and the association of the gene of interest in high-grade serous ovarian cancer progression.
Our study points out new factors of clinical and functional importance in ovarian cancer. We show that a discovered gene of interest affects a known oncogenic signaling pathway, and present that a gene of interest is associated with the common and highly malignant high-grade serous subtype, and poor patient survival.
Authors: Elina A Pietilä1, Lidia Moyano-Galceran2, Joonas Jukonen1, Laura Lehtinen3, S Pauliina Turunen2, Tomas Santos Martins1, Ville Rantanen4, Sanaz Jamalzadeh4, Antti Häkkinen4, Kaiyang Zhang4, Ping Chen4, Tarja Lamminen3, Katja Kaipio3, Johanna Hynninen5, Sakari Hietanen5, Seija Grénman5, Rainer Lehtonen4, Sampsa Hautaniemi4, Olli M Carpén3,6, Kaisa Lehti1,2
Author affiliations: ¹Research Programs Unit, Individualized Drug Therapy, University of Helsinki; ²Department of Microbiology, Tumor and Cell Biology, Karolinska Institutet; ³Department of Pathology, University of Turku and Turku University Hospital Turku; 4Research Programs Unit, Systems Oncology, University of Helsinki; 5Department of Obstetrics and Gynecology, Turku University Hospital, University of Turku; 6Department of Pathology, University of Helsinki and HUSLAB, Helsinki University Hospital; Auria Biobank, University of Turku and Turku University Hospital.
Extracellular matrix (ECM) interactions contribute to cancer metastasis and chemotherapy resistance by regulating invasive cancer growth and apoptosis evasion. Accumulating evidence show specific ECM protein signatures to be associated with higher disease score and poor survival in high-grade serous ovarian cancer (HGSOC). However, the key ECM pathways activated during cancer evolution, including tumor metastasis and tissue responses to chemotherapy, remain to be systematically identified. Combining unbiased transcriptomic analysis of HGSOC tissues with distinct anatomical location and functional ECM screen for platinum-treatment responses, we show how the tumor microenvironment evolves with disease progression and how specific stromal ECM components, including fibronectin, promote HGSOC motility, invasion and resistance to cisplatin treatment. Importantly, we observed these specific ECM proteins to be expressed already at the omental micrometastases, as shown by immunohistochemistry of a longitudinally collected patient cohort. Further, using clinical TCGA data together with loss-of-function screen we identified key integrin subunits driving cancer cell invasion in 3D collagen. These results underline the importance of cell-matrix communication in invasion and chemotherapy response, revealing important promoters of aggressiveness and chemotherapy resistance in the metastatic HGSOC lesions.
Authors: Maria Lindqvist
Author affiliations: Karolinska Institutet, Sweden.
Background: We had earlier shown that Wnt7a which is a ligand for Vangl2 receptor effects apoptosis and migration of epithelial cells. The purpose of current study is to investigate if Vangl2 is effected in wound healing and if this signaling pathway is effected when anesthetics are used.
Aims: 1. Analyzing closer which molecules mediate Vangl2 signalings effect on apoptosis by investigating caspase-8 singling and investigate if this signaling pathway is intervened when anesthetics are applied. 2. Analyzing closer which molecules mediate Vangl2 signalings effect on migration.
Methods: HeLa-cells were cultured on coverslips in 6-well plates until they became 80% confluent. They were transfected with a commercially bought Vangl2-GFP construct for 24 or 48 hours while the controls were transfected with GFP. Gene silencing studies will be done by CRISPR/ CAS9. We further plan to simulate our studies with the help of bioinformatic programs.
Conclusion: We conclude that Vangl2 overexpression effects apoptosis and hypothezise that this is possibly done via caspase-8 signaling. We also suggest that the rearrangement of p53 and Th17 expression obtained in our preliminary experiments is a result of actin translocation from the cytoplasm into the nucleus and is probably mediated by micro-RNAs/long non-coding RNAs and possibly interfered when anesthetics are applied. We further suggest that the delay in differentiation is a result of Vangl2 signalings effect on the expression of Cx43.
Authors: Maren Laasik, MDa,b, Johanna Hynninen MD, PhDa,b, Marko Seppänen Adj prof, MD, PhDa,c , Sakari Hietanen, Adj prof, MD, PhD a,b,d,
Author affiliations: a University of Turku; b Turku University Hospital, Department of Obstetrics and Gynecology; c Turku University Hospital, Department of Clinical Physiology and Nuclear Medicine and PET; d FICAN West, Finland.
Although epithelial ovarian cancer (EOC) is initially chemosensitive, most of the women experience multiple and finally fatal chemoresistant relapses. One of the most important drives of chemoresistance is hypoxia, i.e., subnormal levels of tissue oxygenation. Tumor hypoxia can be evaluated preoperatively with PET imaging using a novel tracer EF5. The feasibility of EF5 PET/CT-imaging in intra-abdominal tumors such as EOC with widespread carcinosis has not been established. Our consortium is first in the world to study ovarian cancer and hypoxia imaging in real clinical context.
In our prospective study design, we aim to locate the intra-abdominal hypoxic tumor areas with preoperative EF5 PET/CT imaging and with targeted surgical sampling in 40 patients with EOC, precisely obtain hypoxic and potentially chemoresistant cancer tissue for our analyses. We also study how hypoxic tumor areas respond to neoadjuvant chemotherapy. We aim to assess DDR and bioenergetic profile, compare the sequencing data from hypoxic tumors and circulating tumor (ct) DNA to assess the mutation profile of circulating tumor cells and its’ predictive value to patient prognosis. The results from this project will advance cancer research and be used in HGSOC patient management.
Authors: Pia Roering1, Vanina Dahlström-Heuser1, Swapnil Potdar2, Johanna Hynninen3, Seija Grénman3, Krister Wennerberg4, Olli Carpén1 and Katja Kaipio1
Author affiliations: 1) Research Center for Cancer, Institute of Biomedicine, University of Turku, Turku, Finland. 2) Institute for Molecular Medicine, High Throughput Biomedicine Unit (HTB), University of Helsinki, Helsinki, Finland. 3) Department of Obstetrics and Gynecology, University of Turku and Turku University Hospital, Turku, Finland. 4) Biotech Research and Innovation Centre (BRIC), University of Copenhagen, Copenhagen, Denmark.
High-grade serous ovarian cancer (HGS-OvCa) is by far the most common and aggressive subtype of ovarian cancer. The standard therapy for HGS-OvCa patients is platinum-taxane chemotherapy together with surgery. While most tumors initially respond to chemotherapy, treatment resistance eventually occurs in 80-90% of the cases. Reasons for relapse and treatment failure are currently largely unknown and treatment options for patients with platinum-resistant disease are limited. Better means to evaluate the efficacy of drug combinations and to suggest alternative combinations in chemo-resistant stage would have a major impact.
We established and characterized primary HGS-OvCa cell lines originated from patient samples to be used as model to get more reliable tools for investigating drug sensitivity and resistance. To identify novel drug compounds or combinations for treatment, high-throughput screening of 306 drugs was performed. Among several promising compounds, we identified an inhibitor (AZD1775) targeting the G2 checkpoint molecule Wee1 with cytotoxic efficacy. It was selected for further in vitro studies.
Our investigation showed that Wee1 inhibition significantly reduced cell proliferation, migration and invasion in all tested primary and commercial ovarian cancer cell lines. Wee1 inhibition also had an effect on the cell cycle and therefore cell cycle inhibited cells cannot proceed into mitosis. This is most likely due to the mitotic catastrophe caused by difficulties with tubulin arrangement in the mitotic spindle. Same tubulin rearrangement problems most likely had an effect on cell motility and ability to invade, which are important processes in cancer progression. Taken together, our results suggest that Wee1 inhibitor is a promising compound for ovarian cancer treatment.
Authors: Domenico Tierno1,2, Giorgio Stanta1, Vincenzo Canzonieri1,3 and Serena Bonin1
Author affiliations: 1: Department of Medical Sciences-University of Trieste- Cattinara Hospital- Trieste, Italy. 2: Doctorate of Nanotechnology, University of Trieste, Trieste, Italy. 3: CRO- Centro di Riferimento Oncologico- Aviano, Italy.
The high mortality of high-grade serous ovarian cancer (HGSOC) patients is often associated to the lack of early diagnosis and to the resistance to chemotherapy drugs commonly used against this tumor: the cis-platin. Biomechanical properties of ovarian cancer cells can provide new insights on HGSOC. Analysis via single-cell of different ovarian cancer cell lines by atomic force microscope (AFM) have shown significant differences in stiffness and in intrinsic distribution of single cells stiffness, with higher variability in HGSOC cell lines (TYKNU = 0.294 ± 0.184 kPa , TYKNU CpR = 0.516 ± 0.231 kPa, OVCAR4 = 1.093 ± 0.510 kPa) than low-grade serous (SKOV3 = 0.769 ± 0.517 kPa , HEY = 0.893 ± 0.677 kPa) and endometrioid (IGROV1 = 0.784 ± 0.384 kPa) ones. Moreover, cis-platin resistance cell line (TYKNU CpR) resulted significantly harder than the corresponding cis-platin sensible ones (TYKNU). These results show the possible use of biomechanical properties to characterize ovarian cancer cells.
Authors: Alejandra Cervera 1*, Heidi Rausio2*, Noora Andersson1, Tiia Kähkönen2, Gabriele Partel1, Johanna Hynninen3, Rainer Lehtonen1, Olli Carpén1, Sampsa Hautaniemi1, Kaisa Huhtinen2
Author affiliations: 1 University of Helsinki, Helsinki, Finland; 2 University of Turku, Turku, Finland; 3 Turku University Hospital, Turku, Finland. * Equal contribution
Background: High-grade serous ovarian cancer (HGSOC) is characterized by excessive chromosomal instability. However, only few recurrent gene fusions (CDKN2D-WDFY2 and BCAM-ALK2) are currently known. Identification of novel gene fusions provides novel diagnostic markers and identifies potential therapeutic targets.
Methods: Whole genome and RNA sequencing data from 125 tumors and primary cell lines from 39 HGSOC patients were analyzed using a novel computational pipeline, which integrates several fusion detection tools, reconstruction and visualization of chimeric transcripts and filtering by functional annotations. After prioritization, the top fusions were validated with Sanger sequencing and by RNAs fluorescent in situ hybridization. Further validation using a Rapid Amplification of cDNA ends method including 49 new RNA samples from 20 patients is ongoing.
Results: Our pipeline resulted in 394 fusion events that were further prioritized. Out of 11 selected fusions 10 were successfully validated with Sanger sequencing. Three of the validated fusions target PI3K-AKT-mTOR signaling cascade that is constitutively hyper-activated in 40-50% of HGSOC and contributes to tumor growth and chemoresistance. The RNA-ISH localized fusion expression specifically in the cancer cells in clinical tumor specimens.
Conclusions: We have identified and validated PI3K-AKT-mTOR as a recurrent target of fusion events in HGSOC. These fusions and the pathways have high potential as clinical targets.
Authors: Liina Salminen1, Nimrah Nadeem2, Anne Lone Rolfsen3, Anne Dørum3, Teemu D. Laajala4,5, Seija Grènman1, Sakari Hietanen1, Taija Heinosalo6, Antti Perheentupa1,6, Matti Poutanen6, Nils Bolstad7, Olli Carpén8,9, Urpo Lamminmäki2, Kim Pettersson2, Kamlesh Gidwani2, Johanna Hynninen1, Kaisa Huhtinen8.
Author affiliations: 1Department of Obstetrics and Gynecology, Turku University Hospital and University of Turku, Turku, Finland; 2Department of Biochemistry/Biotechnology, University of Turku, Turku, Finland. 3Department of Gynecologic Oncology, Radiumhospital, Oslo University Hospital, Oslo, Norway. 4Department of Mathematics and Statistics, University of Turku, Turku, Finland. 5Department of Pharmacology, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA. 6Institute of Biomedicine, Research Centre for Integrative Physiology and Pharmacology, University of Turku, Turku, Finland. 7Department of Medical Biochemistry, Oslo University Hospital, Oslo, Norway. 8Institute of Biomedicine, Research Center for Cancer, Infections and Immunity, Department of Pathology, University of Turku and Turku University Hospital Turku, Finland. 9Department of Pathology and Genome Scale Biology Research Program, University of Helsinki and Helsinki University Hospital, Helsinki, Finland.
Background: The cancer antigen 125 (CA125) immunoassay (IA) does not detect early stage epithelial ovarian cancer (EOC) with the precision needed in clinical practice. In recent studies, glycoforms of CA125 have shown potential as biomarkers in EOC. Here, we assessed the diagnostic abilities of two recently developed CA125 glycoform assays for patients with a pelvic mass. Special attention was drawn at early stage EOC (stage I and II) and patients with marginally elevated conventionally measured CA125 levels.
Methods: Our study population contained 617 patients diagnosed with EOC, benign ovarian tumors and endometriosis. Early stage EOC was diagnosed in 62 patients and marginally elevated serum levels of conventionally measured CA125 was detected in 195 patients, regardless of disease stage. Preoperative serum levels of conventionally measured CA125 and its glycoforms (CA125-MGL and CA125-STn) were determined.
Results: The CA125-MGL assay identified early stage EOC significantly better than the conventional CA125-IA (65% vs 48% sensitivity at a fixed specificity of 90%, p < 0.0001). Further, the combined glycoform assays detected stage I EOC in a subgroup of patients with marginally elevated conventionally measured CA125.
Conclusions: The results indicate that the CA125 glycoform assays markedly improve the performance of the conventional CA125-IA in the diagnosis of early stage EOC. This result is especially valuable when CA125 is marginally elevated.
Authors: Barun Pradhan1, Kaiyang Zhang1, Richard Badge2, Liisa Kauppi1.
Author affiliations: 1University of Helsinki, Systems Oncology research program (ONCOSYS), Helsinki, Finland. 2University of Leicester, Department of Genetics, Leicester, United Kingdom.
Long Interspersed Element-1 (LINE-1) is a family of retrotransposon that makes up to 17% of the human genome, and are the only autonomously active mobile genetic element present in humans. A full-length intact LINE-1 move by a copy-and-paste mechanism, transcribing first into an RNA intermediate, which then eventually integrates into the genome. While epigenetically silenced in normal cells, they are active in cancer cells and can (a) cause deleterious insertions in tumor suppressor genes, (b) deregulate expression of driver genes and/or (c) cause genomic instability. It is challenging to detect novel LINE-1 insertions, as there are more than 500,000 copies of LINE-1 already present in the human genome. However, only about 5,000 LINE-1 copies are full-length and have an intact promoter, out of which only 150 copies are retrotransposition-competent (RC). To study the role of LINE-1 retrotransposons in high-grade serous ovarian carcinoma (HGSOC), we first analyzed RNA-sequencing data from 21 HGSOC samples from 15 patients to identify actively transcribing RC LINE-1 loci that can potentially retrotranspose from one genomic location to another. We found 11 RC LINE-1 loci that were transcriptionally active in at least one sample. We have designed a PCR-based method to detect de novo LINE-1 insertions stemming from three of the most actively transcribing RC LINE-1 loci. We now aim to catalog these LINE-1 insertions in HGSOC patient samples and study their functional relevance.
Authors: Emilia Kozlowska1, Tuulia Valius2, Anniina Färkkilä1, Johanna Hynninen2, Sakari Hietanen2, Marek Kimmel3, Sampsa Hautaniemi1
Author affiliations: 1University of Helsinki, Finland; 2Turku University Hospital, Finland; 3Rice University, US.
A major issue in treatment of advanced ovarian cancer is resistance to standard treatment. Current standard-of-care in ovarian cancer includes platinum-based chemotherapy. Even thought, majority of patients respond very well to platinum-taxane, most of them will relapse within few months after end of treatment. Thus, here we suggest application of mathematical modeling ad simulation approach to find novel drug combination in ovarian cancer, as well as, optimal drug scheduling for platinum-taxane chemotherapy.
Thus, here we first propose application of mathematical modeling to identify effective targeted therapies which could be combined with platinum-based chemotherapy. We developed virtual clinical trials simulation framework based on branching process model to test effectiveness of targeted therapy in rapid and +inexpensive manner. Our results show that drugs that resensitize chemoresistant cells are superior to those aimed at triggering apoptosis or increasing the bioavailability of platinum.
Ovarian cancer cases is heterogenous group of patients which differ in terms of, among other things, tumor load and response to chemotherapy. Thus, we suggest that for each subset of ovarian cancer patients, drug schedule should be adjusted based on tumor properties. We tested this hypothesis by performing mathematical model simulation of heterogenous tumor growth combined with model of pharmacokinetics of cisplatin.
Authors: Ozan Ozisik1,2, Chiara Facciotto3, Antti Häkkinen3, Kaiyang Zhang3, Manuela Tumiati3, Sampsa Hautaniemi3, Benno Schwikowski1.
Author affiliations: 1Yildiz Technical University, Turkey; 2Institut Pasteur, France; 3University of Helsinki, Finland.
We aimed to predict platinum resistance in High grade serous ovarian cancer (HGSOC) patients from mRNA measurements. Based on the known correlation between the activity of the Homologous Recombination (HR) pathway and primary platinum sensitivity [Tumiati et al., Clin Cancer Res 24 (18), 2018], we hypothesized that HR- and Fanconi Anemia (FA)-related genes could be used to classify new patients as either resistant or sensitive.
We performed deconvolution on HGSOC mRNA expression data obtained from HERCULES and TCGA consortia and used the epithelial ovarian cancer component to evaluate our hypothesis. In the HERCULES dataset, we had suitable data from 7 resistant and 17 sensitive patients, while TCGA provided data from 35 resistant and 35 sensitive patients. We built our classifier using genes from KEGG's HR and FA pathways and the Nearest Centroid Classification method. In the HERCULES dataset, the balanced accuracy score was 0.791, while it was 0.586 in the TCGA dataset.
We explored whether any of the HR+FA genes differed between resistant and sensitive patients in the HERCULES dataset. We noticed that ABRAXAS1 significantly differed (padj<0.05 after Benjamini-Hochberg correction, Mann-Whitney U test). Our subsequent accuracy estimate of an ABRAXAS1-only classifier was 0.842. We also found a subgroup of HR-related genes (BRCA1, ABRAXAS1, FANCC, PMS2, RAD50) that led to an accuracy estimate of 0.964.
Authors: Chiara Facciotto1, Antti Häkkinen1, Julia Casado1, Ville Rantanen1, Jaana Oikkonen1, Veli-Matti Isoviita1, Rainer Lehtonen1, Sampsa Hautaniemi1
Author affiliations: 1Research program in Systems Oncoloty, Faculty of Medicine, University of Helsinki.
Platinum-based chemotherapy is a cornerstone treatment for many solid tumor cancers, such as High-Grade Serous Ovarian Cancer (HGSOC). Mechanisms causing resistance to platinum are complex and multi-variate, and not fully understood to date. In order to identify genes and pathways involved in platinum resistance, we collected gene expression, DNA methylation, and copy number alterations (CNAs) of over 900 treatment-naive samples from 12 different solid tumor cancers in The Cancer Genome Atlas (TCGA). We then performed a multi-omics data integration and identified 322 genes whose expression is regulated by DNA methylation and CNAs (p-value <= 0.05 and FDR <= 0.01). Kaplan-Meier analysis of platinum free interval (PFI) resulted in 57 genes having significant survival association (p-value <= 0.05) in TCGA. We then prioritized them based on how stably represented they are across cancer types, identifying, among others, genes SP140, BCAS3, EFNB2, B4GALT2, CD52, and MRPL48, which have been reported to be related in drug resistance in cancers, and ZNF236, TSTA3, JMY, and BPHL that are novel candidates for driving platinum resistance.
Authors: Juha Koiranen1, Antti Häkkinen1, Julia Casado1, Rainer Lehtonen1, Sampsa Hautaniemi1
Author affiliations: 1 Research Program in Systems Oncology, Research Programs Unit, Faculty of Medicine, University of Helsinki.
Single-cell measurements allow profiling 1,000s of markers in 100,000s of phenotypes present in a tissue sample. As the phenotypic landscapes are high-dimensional, intricate, and yet poorly characterized, nonparameric dimensionality reduction is often a first step in gaining insight on these novel data. Few effective methods like the popular t-SNE exist, and their applicability is costly. Currently, t-SNE scales poorly to the ever increasing dimension and volume; parameter tuning requires experimentation; and it is unknown how well the projection captures the underlying data. Consequently, much information remains unexploited and less frequent phenotypes are inadvertently lost.
We improve t-SNE for large-scale analysis while retaining its attractive properties: we implemented a quasi-Newton solver, which can converge quadratically (vs. linearly), allowing optimization in 100 vs. 10,000 iterations; added automatic bandwidth selection, which mitigates the need for parameter tuning; and propose an information-theoretic quality metric, which quantifies how faithfully the projection represents the original data. We use human bone marrow cytometry data, handwritten digits, and cytometry data from ovarian cancer patient tissues to show that the throughput is practically improved by orders of magnitude for various applications. As a result, our improvements permit analysis of complete single-cell datasets, revealing previously undiscovered phenotypic structures.
Authors: Antti Häkkinen1, Kaiyang Zhang1, Amjad Alkodsi1, Anna Vähärautio1, Noora Andersson2, Olli Carpén1,2,3, Rainer Lehtonen1, Sampsa Hautaniemi1
Author affiliations: 1 Research Program in Systems Oncology, Research Programs Unit, Faculty of Medicine, University of Helsinki; 2 Department of Pathology, University of Helsinki and HUSLAB, Helsinki University Hospital; 3 Department of Pathology, Institute of Biomedicine, University of Turku and Turku University Hospital, Finland.
Patient-derived tumor tissue samples consist of various cell types: in addition to the cancer cells, the tissue often contains various stromal and immune cells. Some samples feature only 30% of cancer cells (even select TCGA samples often contain only 60%), hindering sample comparison especially when comparing samples before and after chemotherapy. At best, the findings in subsequent analyses are diluted, due to the presence of an unchanging component, while at worst the treatment systematically affects the composition, and the findings cannot be associated with phenotypic changes.
To counter this, we developed a method that estimates both composition and expression of bulk RNA samples guided by single-cell RNA-seq profiles. Unlike alternatives, our method does not require replicates, manually curated reference profiles, or estimates of the composition from other sources (e.g. IHC). Our method also adapts to the heterogeneity in each cell type, in unmatched patients, and accommodates for unseen cell types. Further, we show how the putative cell types composing the samples can be automatically discovered. Using our method, we show that the composition of high-grade serous ovarian cancer bulk samples is confounded by the treatment phase and tissue site but that the composition may not be a strong prognostic factor. Instead, by applying our method, we can enrich cell type specific expression, and construct stronger prognostic factors for predicting the patient response.
Authors: Sanaz Jamalzadeh1, Kaiyang Zhang1, Antii Häkkinen1, Kaisa Huhtinen2, Johanna Hynninen3, Olli Carpen2,4, 5, Sakari Hietanen3, Rainer Lehtonen1, Sampsa Hautaniemi1
Author affiliations: 1Research Programs Unit, Systems Oncology, Department of Biochemistry and Developmental Biology, Faculty of Medicine, University of Helsinki, Helsinki, Finland. 2Department of Pathology, University of Turku and Turku University Hospital, Turku, Finland. 3Department of Obstetrics and Gynecology, University of Turku and Turku University Hospital, Turku, Finland. 4Department of Pathology, University of Helsinki and HUSLAB, Helsinki University Hospital, Helsinki, Finland. 5Auria Biobank, University of Turku and Turku University Hospital, Turku, Finland.
Background: High grade serous ovarian cancer (HGSOC) is an aggressive disease treated with platinum and taxane combination therapy. Although most patients at the advanced stage of the disease initially respond to it, the majority suffers relapse within 18 months. Therefore, identification of leading signatures is crucial in limiting such platinum resistance in HGSOC patients.
Aim: Comparing RNA-sequencing data of HGSOC patients with poor and good response to chemotherapy in order to find dominant mechanisms in platinum resistance deals with differences between anatomical locations. To this end, by adjusting for anatomical locations' effect in decomposed bulk RNA-Seq expression data, we would be able to find differentially expressed genes (DEGenes) between poor and good responders as the key genes involved in platinum resistance. Besides, searching for association between the most significant DEGenes and progression free survival of patients with ovarian cancer in other cohorts, such as TCGA, will validate the prognostic signatures in HGSOC more precisely.
Authors: Jun Dai1§, Erdogan Pekcan Erkan1§, Kaiyang Zhang1§, Katja Kaipio2, Tarja Lamminen2, Naziha Mansuri2, Lasse Suominen1, Kaisa Huhtinen2, Sakari Hietanen3, Johanna Hynninen3, Seija Grénman3, Olli Carpén4, Sampsa Hautaniemi1, Anna Vähärautio1*
Affiliations: 1Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Helsinki, Finland. 2Institute of Biomedicine; Research Center for Cancer, Infections and Immunity, University of Turku, Turku, Finland. 3Obstetrics and Gynecology, University of Turku and Turku University Hospital, Turku, Finland. 4Helsinki Biobank, University of Helsinki and Helsinki University Hospital, Helsinki, Finland.
§Equal contribution *Corresponding Author, anna.vaharautio (at) helsinki.fi
To meet the metabolic demands of high growth rates, tumor cells must overcome nutrient and oxygen deprivation. These environmental stress factors may promote quiescence, allowing tumor cells to evade antiproliferative therapies. The genetic makeup of the tumor also modulates the probability of a quiescent state. A good example of this phenomenon is high-grade serous ovarian cancer (HGSOC), where BRCA1/2 mutations drive hyperproliferation, making tumors initially very responsive to platinum therapy. Still, a small subpopulation survives and relapses after dormancy, often responsive to platinum therapy, suggesting chemotherapy may promote selection by cell state plasticity.
To understand the fundamental mechanisms of chemoresponse, we performed single-cell RNA sequencing on 49 clinical HGSOC specimens. Patients with no visible residual disease had higher E2F1 regulon activity linked to BRCAness, proliferation, and metabolism in their tumors, whereas patients with residual disease had higher FOXO3 regulon activity linked to stress resistance, stemness, and quiescence in their tumors. Moreover, our preliminary data demonstrate that suppression of BRCA1 expression reduces FOXO3 and BCL2L1 mRNA expression ex vivo, suggesting a dynamic interplay between these antagonistic regulons, and supporting the inhibitory role of BRCAness in quiescence. We will further explore the balance between genetic and environmental factors to identify vulnerabilities to tackle chemoresistance in HGSOC.
Authors: Kari Lavikka, Jaana Oikkonen, Rainer Lehtonen, Sampsa Hautaniemi.
Author affiliations: Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Finland.
Visualization is a cornerstone of data exploration. Cancer genomes exhibit various genomic alterations such as copy number changes, point mutations, and changes in methylation. However, current state-of-the-art tools are inadequate for visualizing these multiple dimensions interactively, particularly when hundreds of samples are being studied. Furthermore, the raw data is often augmented with annotations from various databases or scoring algorithms, thus, their full utilization requires considerable flexibility from the visualization tool.
GenomeSpy takes on the challenge by three themes:
- Sophisticated sample sorting and filtering based on either sample-specific attributes (e.g. progression free interval) or genomic data (e.g. copy number)
- A grammar for building layered visualizations using primitive graphical marks; instead of displaying specific file formats in a specific way, the user can freely map attributes of the data to different visual channels such as color or height
- Fluid interactions that allow the user to navigate around the genome and samples effortlessly. Smooth animations allow the user to maintain the context and stay in the flow of exploration
The tool is used in the HERCULES project to visualize copy number changes, loss of heterozygosity, and point mutations in ovarian cancer samples.
Authors: Nil Campamà Sanz1, Jaana Oikkonen1, Antti Häkkinen1, Amjad Alkodsi1, Olli Carpén1,2, Kaisa Huhtinen3, Sakari Hietanen4, Seija E. Grénman4, Rainer Lehtonen1 and Sampsa Hautaniemi1
Author affiliations: 1 Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Finland. 2 Department of Pathology and Forensic Medicine, Institute of Biomedicine, University of Turku, Turku, Finland. 3 Institute of Biomedicine, Research Center for Cancer, Infections and Immunity, University of Turku, Turku, Finland. 4 Department of Obstetrics and Gynecology, University of Turku, Turku University Hospital, Turku, Finland.
High-grade serous ovarian cancer (HGSOC) is a heterogeneous cancer where majority of the patients will eventually relapse. HGSOC originates mainly from fallopian tube from where it spreads to ovaries and further to nearby tissues. We performed phylogenetic analyses of 88 samples from 18 patients to identify pattern of evolution. Aim was to understand the origin of the resistance mechanisms.
We employed whole genome sequencing in 3-11 tumor tissue samples per patient. Patients were part of the HERCULES project. Samples were gathered from diagnosis (debulking surgery or laparoscopy) and interval debulking surgery (neoadjuvant chemotherapy treated patients) and/or relapse.
Phylogenetic analysis demonstrated complex spreading of the tumor, and many samples revealed subclones originated from multiple cells. High heterogeneity was detected especially between samples from different tissues. Samples from ovaries revealed highest heterogeneity and early branching. Patients with high intra-sample heterogeneity had poor outcome based on platinum free interval (PFI). In a few cases, who had platinum sensitive disease during primary therapy, the relapse samples showed higher heterogeneity when patients were no longer platinum sensitive. Overall, in six cases we detected major change in the subclonal composition between primary and later samples from the same tissue, whereas only two showed similar composition. The spatial and temporal heterogeneity complicate the efforts to overcome resistance in HGSOC.
Authors: Jaana Oikkonen1, Kaiyang Zhang1, Liina Salminen2, Ingrid Schulman1, Kari Lavikka1, Noora Andersson1, Erika Ojanperä1, Sakari Hietanen2, Seija Grénman2, Rainer Lehtonen1, Kaisa Huhtinen3, Olli Carpén1,4, Johanna Hynninen2, Anniina Färkkilä1,5 and Sampsa Hautaniemi1
Author affiliations: 1 Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Finland. 2 Department of Obstetrics and Gynecology, University of Turku, Turku University Hospital, Turku, Finland. 3 Institute of Biomedicine, Research Center for Cancer, Infections and Immunity, University of Turku, Turku, Finland. 4 Department of Pathology and Forensic Medicine, Institute of Biomedicine, University of Turku, Turku, Finland. 5 Department of Radiation Oncology, Dana-Farber Cancer Institute, Harvard Medical School, USA.
Circulating tumor DNA (ctDNA) has been proposed as a “real-time” biomarker to identify information about tumor in a non-invasive manner. Using ctDNA, it is possible to detect tumor mutations and copy number alterations (CNAs) to follow treatment response and tumor progression during and after primary therapy. Our aim was to translate ctDNA profiling into clinical benefit in patients with high-grade serous ovarian cancer (HGSOC).
We sequenced 100 longitudinal plasma samples from 15 patients with a comprehensive cancer-specific panel of over 500 genes.
We identified several actionable mutations and CNAs. In one chemo-resistant patient, detection of ERBB2 amplification from pre-treatment ctDNA resulted to ctDNA-guided treatment including trastuzumab during tumor progression, which yielded tumor shrinkage. In two other patients, mTOR pathway activation was predicted based on mutations detected in ctDNA suggesting benefit from mTOR inhibitors in case of disease progression.
Response to chemotherapy was detected from ctDNA already after first cycles. Poor-responding patients with platinum free interval (PFI) less than 12 months showed failure to drop ctDNA level after start of chemotherapy and higher number of detected mutations in diagnosis. Relapse was clearly detectable from ctDNA in 80% of the cases. Early prognosis prediction in combination with identification of clinically relevant variants can allow a window of opportunity to treat poor-prognosis patients even before relapse.
Authors: Luca Pasquini1, Julia Casado2, Katja Kaipio3, Eleonora Petrucci1*, Oskari Lehtonen2, Tarja Lamminen3, Kirsi Toivanen3, Alessandra Boe1, Olli Càrpen3, Sampsa Hautaniemi2, Mauro Biffoni1*.
Author affiliations: 1Istituto Superiore di Sanità, Core Facilities, Rome, Italy. 1*Istituto Superiore di Sanità, Department of Oncology and Molecular Medicine; Rome, Italy. 2University of Helsinki, Hautaniemi Lab, Finland. 3University of Turku, Carpen Lab, Finland.
High-grade serous ovarian cancer (HGSOC) is the most common histological subtype of ovarian carcinoma characterized by initial response to platinum-based standard treatments followed by relapse and increasing chemoresistance. Platinum resistance in HGSOC likely originates from high intra-tumoral heterogeneity, which is related to a poor prognosis. Identification of tumor subpopulations responsible for therapeutic failure will help to drive new personal treatments bypassing chemoresistance.
In this preliminary study, we characterized by mass cytometry 19 fresh HGSOC samples collected in the HERCULES project from different anatomical sites before and after chemotherapy. We employed a pilot panel including 27 antibodies against surface/intracellular antigens and phosphoproteins, to discriminate cancer cells from infiltrating cells. We developed computational framework to analyze data, which automatizes state of the art methods with interactive visualization to focus key signatures that can describe the cell population of each acquired sample.
This approach could discriminate three main cell populations outlining patient-specific signatures within the tumor compartment, as well as patient-specific tumor cell phenotypes. The dominant cell populations were characterized by epithelial cells, as well as significant immunosuppressive, stem-like phenotype or potential EMT traits. The ability to stratify patients based on the tumor cell composition could be useful to facilitate precision medicine treatments.
Authors: Miikka Kilkkilä1, Julia Casado1, Connor Jacobson2, Antti Häkkinen1, Zoltan Maliga, Peter Sorger2, Sampsa Hautaniemi1, Anna Vähärautio1, Anniina Färkkilä1,2.
Author affiliations: 1Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Finland. 2Harvard Medical School, Boston, US.
A key event in the progression of High Grade Serous Ovarian Cancer (HGSOC) is the establishment of an immunosuppressive environment that prevents the immune system from effectively attacking the cancer cells. The determinants of immune suppressive cell types and pathways in HGSOC are poorly characterized. In this study we utilized tissue Cyclic Immunofluorescence (tCycIF), a novel highly multiplexed single-cell proteomics platform that allows quantification up to 60 antigens at single cell resolution from a single formalin-fixed paraffin embedded (FFPE) tissue slide. We compared the detected cell types and proportions to the results from bulkRNA-seq deconvolution and scRNA-seq analysis.
We studied 4 samples that had both tCycIF and bulkRNA-seq data from the HERCULES cohort and found a good correlation with the tumor compositions in the samples. Some immune cells, such as CD8+T-cells, however, have large compositions in tCycIF samples while being totally absent in RNA-seq data. The improvement of novel single-cell immunophenotyping methods is critical to develop more effective immunotherapies to improve the outcomes and survival of patients with HGSOC.
Authors: Oskari Lehtonen1, Julia Casado1, Katja Kaipio3, Luca Pasquini2, Eleonora Petrucci2, Tarja Lamminen3, Kirsi Toivanen3, Mauro Biffoni2, Olli Càrpen3, Sampsa Hautaniemi1
Author affiliations: 1Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Finland. 2Instituto Superiore di Sanita, Department of Haematology, Rome, Italy. 3Department of pathology, University of Turku, Finland.
New technological advances in flow and mass cytometry (CyTOF) have made it possible to detect expression levels of dozens of protein markers in millions of individual cells, enabling scientists to study heterogenous cell populations at a single cell level. The high dimensionality of these data has led to an ever-increasing number of automated analysis tools to visualize the data. These analysis tools usually have very similar structure with small variations to their dimensionality reduction and unsupervised clustering methods. For each iteration on the analysis to interpretation cycle, cytometrists require new data visualizations with custom scripts on varying programming languages and libraries, which can make many of these methods unattractive especially to new users.
Here we present Cyto, an open source solution developed in the HERCULES project, built on top of the workflow engine Anduril. Cyto supports state of the art cytometry analysis methods and allows for rapid integration of new emerging methods under the same framework, even when the methods are built with different programming languages. Moreover, Cyto pipelines can be enabled for non-technical users through graphical user interface. In this study we present a flexible CyTOF analysis pipeline with an easy-to-use graphical user interface and a fully interactive result browser to analyze or share the results independent the specific experimental setting. The ability to integrate latest bioinformatics tools with interactive visualization of the data removes the bottleneck from high-dimensionality analysis.
Authors: Kaiyang Zhang1, Antti Häkkinen1, Kaisa Huhtinen2, Johanna Hynninen3, Olli Carpén1,2, Sakari Hietanen3, Rainer Lehtonen1, Sampsa Hautaniemi1
Author affiliations: 1Research Program in Systems Oncology, Faculty of Medicine, University of Helsinki, Helsinki, Finland. 2Department of Pathology and Forensic Medicine, Institute of Biomedicine, University of Turku, Turku, Finland. 3Department of Obstetrics and Gynecology, University of Turku, Turku University Hospital, Turku, Finland.
In cancers transcriptional dysregulation of genes leads to aberrant cell growth and resistance to chemotherapy. Quantification of gene regulatory network (GRN) activities is important for identification of prognostic markers and patient subgroups, that could be used for optimizing treatment strategies to improve clinical outcome.
Although a number of pathway or gene set scoring approaches, such as SPIA, GSEA, AUCell, have been developed and applied to reveal GRN activities, most of them are designed to evaluate only one type of omic data at a time. We present a novel method for inferring GRN activities from multiple types of omic data, in which a GRN is modeled by a Bayesian network and different layers of omics data (e.g. copy number, expression, protein activities, etc.) are encoded by continuous variables. Downstream nodes are assumed to be linearly dependent on their parents, and belief propagation is used for the inference of the activity of each GRN in each sample.
We applied our method to 566 transcription factor (TF) - targets networks in 392 samples from The Cancer Genome Atlas (TCGA) high-grade serous ovarian cancer (HGSOC) cohort and 52 samples from the HERCULES cohort. We identified POU3F3, DDIT3, THRA to be significantly associated with disease free survival in the TCGA cohort and with progression free survival in the HERCULES cohort, suggesting their involvement in the development of chemotherapy resistance in HGSOC.
Authors: Wei Bei1, Jussi Taipale2.
Author affiliations: 1Karolinska Institutet, Sweden; 2University of Cambridge, UK.
It is well known that transcription factors (TFs) play crucial roles in determining cell identity. In order to identify such cell-type specific signature factors, we have established a novel method, active TF identification (ATI), where nuclear protein extracts are incubated with a highly complex DNA library and subjected to the electrophoretic mobility shift assay followed by sequencing. Computational analysis of the enriched sequences allows us to profile and characterize the most active TFs in the original extract.
By applying this method in the HERCULES project to tissues and cell lines from multiple different species, we found that only ~ 10 TFs display high activities in any given cell or tissue type. Within these highly active TFs, there were both housekeeping TFs, which were universally found in all cell types, and specific TFs, which were highly enriched in known factors that determine the fate of the analysed tissue or cell type. Our results indicate that in a given cell type, gene expression may be strongly determined by the selection of the members of a small but strongly active TF set, and as a whole, the gene regulatory logic may be far simpler than what has previously been appreciated. ATI can be used as powerful tool towards identifying essential transcription factors to understand transcriptional networks operating in HGS-OvCa.
Authors: Aleksandr Ianevski1, 2,Alexander Kononov1, Sanna Timonen1, Tero Aittokallio1,2,3, Anil Kumar Giri1*
Author affiliations: 1 Institute for Molecular Medicine Finland (FIMM), University of Helsinki, FI-00290 Helsinki, Finland. 2 Helsinki Institute for Information Technology (HIIT), Aalto University, FI-02150 Espoo, Finland. 3 Department of Mathematics and Statistics, University of Turku, Quantum, FI-20014 Turku, Finland.
Drug combinations are becoming a standard treatment of many diseases due to their capability to tackle drug efficacy and resistance problems. Currently, due to the lack of computational methods tailored to profile synergy, efficacy, and toxicity of drug pairs simultaneously the selection of top hits for further study is often done merely based on synergy, without considering the combination efficacy and toxicity effects, even though these are critical determinants for the clinical success of a therapy. To facilitate the concept of drug combination selection based on integrated, synergy, efficacy and toxicity profiles, we implemented a web-based tool, SynToxProfiler (Synergy-Toxicity-Profiler). When applied to 1960 drug combinations tested in 7 ovarian cancer ALMANAC-NCI cell lines, SynToxProfiler was able to prioritize clinically-established drug combinations (e.g. ìfosfamide and oxaliplatin, decitabine and ifosfamide) as top hits.
Authors: Yevhen Akimov1, Daria R. Bulanova1, Krister Wennerberg1,2, Tero Aittokallio1
Author affiliations: 1 Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki, Finland. 2 Biotech Research and Innovation Centre, University of Copenhagen, Copenhagen, Denmark.
The heterogeneity of HGSOC imposes the need for methods that enable profiling of individual cancer subclones to understand the basis of drug response and the mechanisms of recurrence. We used the DNA barcoding and NGS-based lineage tracing approach to quantify the clonal responses to chemotherapy in ovarian cancer cells. The analysis demonstrated a robust correlation between clonal division rate and the sensitivity to the cytotoxic agents, and provided evidence that the approach can be of use for high-throughput profiling of the clonal drug responses in HGSOC cultures.
To address the accuracy of clonal phenotype quantification via DNA barcoding, we developed an algorithm for reliable calling of differentially represented barcodes in the NGS data. To combine DNA barcoding with the isolation of specific clones for subsequent profiling, we functionalized DNA barcodes to serve both as heritable cellular tags for monitoring clonal dynamics and as recruitment sites for the dCas9-fused transcriptional activators. The barcode-guided CRISPR activation of the puromycin resistance in the specified clones allows for their isolation in a highly controlled fashion. The pilot experiments demonstrated the specificity of the method for the isolation of clones representing ~0.1% of cancer cells population. We are planning to utilize it for isolation and functional characterization of rare chemoresistance-driving HGSOC clones before and after chemotherapy treatment.
Authors: Daria R. Bulanova1, Yevhen Akimov1, Sergei Gordeev1, Manuela Tumiati3, Liisa Kauppi3, Tero Aittokallio1, Krister Wennerberg1,2
Author affiliations: 1 Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki, Finland. 2 Biotech Research and Innovation Centre, University of Copenhagen, Copenhagen, Denmark. 3 Research Programs Unit, Medicum, University of Helsinki, Helsinki, Finland.
HGSOC develops resistance and recurs in 12-18 months after initial response to platinum-based chemotherapy. To identify clinically relevant agents that sensitize HGSOC cells to the therapy, we took an integrated approach that combines CRISPR/Cas9 screening and drug sensitivity testing to explore synergistic effect of carboplatin with approved and investigational cancer drugs.
We identified that in chemo-naïve ovarian cancer cells the genetic knock-out of caseine kinase subutnit alpha 1 (CSNK2A1) and the specific inhibitor of CK2 kinase have strong synergistic effect with carboplatin in a subgroup of patient samples. Further mechanistic analysis revealed that the synergistic interaction depended on the role of CK2 in the DNA repair signaling, and the response to the drug combination can be predicted by the functional assay for homologous recombination repair efficiency. CRISPR knock-out screening on in vitro-developed persister-enriched ovarian cancer cell lines identified activators of ferroptosis and autophagy modulators as persister-specific targets in several cell line models. We demonstrate that our integrated approach enables high-throughput identification of the determinants of carboplatin response in ovarian cancer cells.
Authors: Manuela Tumiati1*, Sakari Hietanen2, Johanna Hynninen2, Elina Pietilä1, Anniina Färkkilä1,3, Amjad Alkodsi1, Yilin Li1, Rainer Lehtonen1, Erdogan Pekcan Erkan1, Minna M. Tuominen1, Kaisa Lehti1,4, Sampsa K. Hautaniemi1, Anna Vähärautio1, Seija Grénman2, Olli Carpén1,5, and Liisa Kauppi1*
Author affiliations: 1ONCOSYS Research Program, University of Helsinki, Helsinki, Finland. 2 Department of Obstetrics and Gynecology, Turku University Hospital, Turku, Finland. 3 Department of Radiation Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA. 4 Department of Microbiology, Tumor and Cell Biology, Karolinska Institutet, Stockholm, Sweden. 5 Department of Pathology, University of Turku and Turku University Hospital, Turku, Finland.
Purpose: Homologous recombination deficiency (HRD) correlates with platinum sensitivity in ovarian cancer patients, which clinically is the most useful predictor of sensitivity to PARPi. For this reason, there is clearly a critical need to identify homologous recombination deficiency (HRD) as early as possible. Currently, HRD is mostly diagnosed by genetic testing, which however fails to identify a large proportion of HR-deficient tumors and predict a patient’s response to chemotherapy, thus we aimed to develop an ex vivo functional HRD detection test that could predict both platinum-sensitivity and patient eligibility to targeted drug treatments.
Methods: We obtained a functional HR score by quantifying homologous recombination (HR) repair after ionizing radiation-induced DNA damage in primary ovarian cancer cells and tissues. Samples clustered in 3 categories: HR-deficient, HR-low and HR-proficient. We analysed the HR score association with platinum sensitivity and treatment response, platinum-free interval (PFI) and overall survival (OS), and compared it with other clinical parameters. In parallel, we performed whole-genome sequencing and targeted DNA-sequencing of HR genes to assess if functional HRD can be predicted by currently offered genetic screening.
Results: Low HR scores predicted primary platinum sensitivity with high statistical significance (p=0.0103), associated with longer PFI (HR-deficient vs HR-proficient: 531 vs 53 days), and significantly correlated with improved OS (HR score <35 vs ⩾35, hazard ratio=0.08, p=0.0116). At the genomic level, we identified a few unclear mutations in HR genes and the mutational signature associated with HRD, but, overall, genetic screening failed to predict functional HRD.
Conclusions: We developed an ex vivo assay that detects tumor functional HRD (both in FFPE and cultured primary cells) and an HR score able to predict platinum sensitivity, which holds the clinically relevant potential to become the routine companion diagnostic in the management of ovarian cancer patients.