#	Pagina
attuale pagina	/open-h2020/projects/208518/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - TheranOMICS (Integration of high resolution -OMICS datasets towards personalized therapy in bladder cancer)

Teaser

Summary

Problem/ issue that was addressed through TheranOMICS Project:
TheranOMICS focuses on improving Bladder Cancer (BC) management. BC is the second most frequent cause of mortality among genitourinary cancers and the costliest malignancy of all to manage. It is classified into non-muscle invasive BC (NMIBC, which is less advanced) and muscle invasive BC (MIBC, more advanced). Invasiveness results in poor prognosis, since more than 50% of patients succumb to their disease within 5 years. Unfortunately, patients with less advanced cancer frequently re-develop cancer and in some cases in advanced forms. Moreover, treatment options are limited for the advanced cancer, resulting in very high mortality. TheranOMICSâ€™ idea was the development of tools that could timely and accurately predict the disease progression. This will support the management of the patients by guiding intervention and on the other hand, will assist the drug development to support the release of novel drugs. The latter is directed towards pharmaceutical industry and is equally important, as currently the efficiency of oncological drugs is low. This problem is attributed to the disease heterogeneity that results in low response rates to the developed drugs. As conventional pathological classification seems not sufficient, the way forward is to integrate available molecular data (-omics features) to define disease classes at the molecular level. From the Innovation and Industrial point of view, by offering companion tools to pharmaceutical industry for stratification of patients into those that are more likely to present with an advanced cancer form and/or those responding to specific treatment, several benefits are offered including the decrease in the overall development cost for development of BC drugs. After the completion of the TheranOMICS project, we can conclude that the recruitment of the Innovation Associate was crucial for the implementation of the above predictive tools.The collaboration resulted in the successful integration of the available -omics data, that Mosaiques diagnostics (Host institution) had acquired and the transformation into clinically meaningful and industrially marketable tools. This was only possible based on the unique background of the Innovation Associate that combined expertise in bioinformatics, data analytics and statistics.

Importance for Society:
TheranOMICS targets to support the running clinical trials and also propose clinical trials that are initiated at an earlier timepoint. The ultimate goal is the improvement of the management of BC patients. Aiming at earlier intervention, there are substantial benefits for patients, such as better survival rate, but also better quality of life for the patients. The need for guidance in treatment is reflected by the fact that the current therapies for BC are not fully efficient and are accompanied by serious side effects. In fact, for less advanced BC, treatment with Bacillus Calmette-Guerin (BCG) is recommended. Unfortunately, there are several side effects and approximately 40% of the patients do not respond. Therefore, as cancer progresses, radical cystectomy is performed along with chemotherapy in cases combined with recently introduced immunotherapy. However, survival in the advanced cases is decreasing dramatically. Collectively, there is a need to improve on the therapies, but also to guide intervention individually to patients that have high probability to respond and thus benefit from the therapy compared to those that do not respond, while for the latter avoiding overtreatment and the associated side effects.

TheranOMICS overall objectives were:
1. The recruitment of a highly skilled Associate to establish multi-parametric, robust molecular profiles which can predict disease outcome and treatment response.
2. The integration of -omics data via bioinformatics tools to reveal molecular features associated with disease outcome.
3. The development of tools for stratifyi

Work performed

TheranOMICS work was performed as planned. To complete the above goals, the Work was organized into two tasks, including a) the development of the predictive tools, b) the training of the Associate. For TheranOMICS project, ethical approval was obtained by the Ethics Committee of Hannover Medical School.

In order to develop the predictive tools, the following process was followed, once ethics approval was obtained. At first, the data were aligned and organized. Different algorithms were investigated and comparatively assessed. The -omics data were analysed and the machine learning models were optimized. As a last step, the Associate together with Mosaiques diagnostics concluded on the final tools and a dissemination strategy was developed, as planned. Regarding the first tool to predict disease outcome, a dataset of omics profiles of BC patients (n=98) was split into a training and test set. 36 features predictive of BC relapse (p-value < 0.1) were further used in the development of a predictive model of BC relapse. A Random Forest model was developed using data from the training set (n = 48) and its performance was assessed in the validation cohort (n = 50). The outcome is a tool that scores the BC patients individually. Patients that score positive have a 5-fold increased probability of developing a disease relapse. A second omics-based tool was developed for prediction of response to BCG treatment. The tool was developed based on omics data available from The Cancer Genome Atlas (TCGA) Research Network/ Pan-Cancer Atlas. A dataset of 265 BC patients originating from the TCGA study was investigated.A subpopulation of treated patients was considered (n=23). 7 proteins were associated with positive response, while 4 indicated a poor response to treatment. The development of the predictive tool was based on Random Forest and a fully functional model with a good accuracy of 66% was finally released.

Regarding the training, two separate approaches were followed: a) a tailored training on the SME\'s infrastructure, database and in-house software and b) a core training provided by the EC. Additional training through a Hands-on Machine Learning course was provided by GOTO Academy and a course for Machine Learning A-Zâ„¢, Python & R (www.udemy.com) was followed online. A core training was provided by the European Commission, including four DACAPO seminars on industrial innovation and business management.

Final results

TheranOMICS resulted in the launch of two predictive tools to assist the development of better drugs for BC. The impact of TheranOMICS is dual. At the clinical level, better stratification of the patients can guide selection of the therapy of choice regarding the available treatment regimens. Moreover, when considering the Innovation- Industrial benefits, the tools are readily available to be offered to the pharmaceutical industry, which opened a new path in the Mosaiques diagnostics pipeline. Although the clinical trials in the field of BC are expanding rapidly, the existing treatment options are still limited and guidance for selecting the most appropriate treatment option per patient is lacking. Within TheranOMICS and through the use of the predictive tools for the stratification of the patients, the approach can impact the development of BC drugs, ultimately positively affecting patient treatment.