PanDrugs is one of the missing links between the potential of biomarkers as predictive tools for treatment outcome and their actual use in clinical settings. Combined with adequate computational infrastructures for data processing and storage, and put in the hands of qualified in-house bioinformaticians, the new platform will guide the selection of therapies from the results of genome-wide studies in cancer disease. ‘Most clinical trials currently evaluate the efficacy and safety of a new drug by analysing its effects on largely unselected populations of patients,’ observes Dr Fatima Al-Shahrour, coordinator of PERSMEDOMICS (Bioinformatics and Integrative Genomics for a Novel Personalized Cancer Therapy) for the Spanish National Cancer Research Centre (CNIO). ‘By using genomic profiles instead within ‘basket trials’, we could bring more accurate diagnosis and treatments based on multi-biomarkers, safer drug prescription, better disease prevention and, consequently, a reduction in healthcare costs.’ Of course, personalised cancer treatment is still in its infancy. Before it can become the norm, hospitals will need quick, reproducible, easy to perform and low cost testing methods. Molecular and clinical data sharing should be promoted, a complete understanding of tumour biology is required, and current therapeutic and pharmacological limitations should be overcome. ‘The latest studies have also shown that personalised medicine requires information from thousands of patients for accurate decision support. Unfortunately, even if all the above-mentioned problems were solved, the computing infrastructures that are currently found at healthcare institutions would still be ill-prepared to efficiently process such volumes of data,’ Dr Al-Shahrour deplores. In this context, the development of fast and cheap technologies linking patient data to other known information is desperately needed. PanDrugs might be part of the solution: By using it, bioinformaticians can analyse and integrate genomic data (mutations, copy number variations or gene expression levels), functional data (protein essentiality), and pharmacological data (sensitivity or resistance to antitumour drugs) – so as to identify actionable molecular alterations. ‘PanDrugs’ goal is to evaluate big data generated by cancer patients’ genomic profiles, according to their biological and clinical relevance and their susceptibility to be pharmacologically-targeted, in order to assist clinical decision making. Additionally, our group has been working on a new drug repositioning methodology to predict sequential treatments in cancer using transcriptional signatures,’ says Dr Al-Shahrour. The PERSMEDOMICS team has already obtained very promising preliminary results. Their tools have notably been used to analyse Pancreatic ductal adenocarcinoma (PDAC) patients sequencing data and prioritise those that may have therapeutic implications. ‘In the context of clinical research projects and clinical trials, PanDrugs has been integrated as a new module in the sequencing analysis pipeline to categorise patient tumours and match them to effective drugs or treatments,’ Dr Al-Shahrour enthuses. ‘So far, we have analysed data from more than 500 patients and this new pipeline has allowed us to identify actionable mutations in nearly half of them.’ One example of this work is a study on 25 patients that consisted in the integration of sequencing data and Patient-derived xenografts (PDXs) mouse models. Using PanDrugs, the team could identify putatively actionable tumour-specific genomic alterations in most cases, whilst experimental testing of candidate treatments in PDXs models helped to select empirical treatments in patients with no actionable mutations. ‘In 2015, we started testing this strategy under an ERC Advanced grant, in a prospective randomised clinical trial on 150 patients with resistant metastatic pancreatic cancer. Our aim was to test the hypothesis that an integrated personalised treatment approach improves survival compared to the conventional treatment strategy,’ says Dr Al-Shahrour. With PanDrugs providing a solution to study molecular profiles in patient populations, public healthcare systems can now focus on gathering the infrastructure, software and expertise required to store, mine, process and analyse such information. PanDrugs is open source, which means that all algorithms and methods are public and freely accessible. The project consortium has already established collaborations with public entities and private companies in the health sector to promote the usage of their tools.
PESMEDOMICS, Pandrugs, cancer, genomics, bioinformatics, big data, personalised medicine, clinical trial, tumour, algorithm