• Director Alexandru FLOARES, PhD, MD
    • University of Medicine and Pharmacy Iuliu Hatieganu Cluj Napoca
    • INCDO-INOE 2000, Research Institute for Analytical Instrumentation, ICIA Subsidiary

Intelligent Systems for Recurrence and Progression Prediction in Superficial Bladder Cancer based on Artificial Intelligence and Microarray Data: tumor mRNA and plasma microRNA – IntelUro

PCCA 2011/2012 –  Biotechnologies –  PN-II-PT-PCCA-2011-3.1-1221

Objective – IntelUro will develop and implement the most accurate (95%-100%) intelligent predictive systems for superficial bladder cancer recurrence and progression, and the bioinformatics methodology and workflow to reach such results. For a new patient, these systems will take as inputs the level of plasma microRNA (non-invasive) or tumor mRNA biomarkers discovered by us in IntelUro, and will predict if the patient will evolve toward recurrence/progression or not. The non-invasive systems are to be preferred except for clinical situations where the invasive surgical procedure cannot be avoided. Thus, the therapy could be personalized for the benefit of the patient with a dramatic decrease of the management costs.

Bladder cancer- Bladder cancer  is the fifth most common neoplasm accounting for about 5-7% of all new diagnosed malignancies in men, and about 2-2.5% in women. In addition to male gender, other risk factors include high age, tobacco smoking and occupational exposure to carcinogens.
From a clinical perspective, bladder cancer has been traditionally subdivided into either superficial (Ta or T1) or invasive (T2, T3, and T4) subtypes. The two types are quite different in their overall characteristics, with superficial disease having long-recurrence free episodes in most cases, while invasive disease often requires multimodal therapies combining surgery as well as chemotherapeutic approaches.

Biomarkers: tools for predicting disease evolution – The fact that bladder cancer is still an exceptionally lethal disease with an annual mortality almost equivalent to its annual incidence has stimulated intense research efforts directed at understanding the underlying molecular mechanisms. Important goals are the detection of cancer at an early stage, determining prognosis and monitoring disease progression and therapeutic response.
Recent experimental results suggest that patients with the same type of cancer often have dissimilar genetic defects in their tumors. These findings explain why patients with the same type of cancer have a different evolution and respond differently to the treatments applied. In consequence, cancer therapy is slowly shifting towards a more personalized approach in which each patient is treated according to the specific defects in the tumor (Veer and Bernards, 2008). Personalized medicine is the ultimate goal of modern cancer treatment and its success depends on the availability of cancer biomarkers (biological indicators) that can be used to guide treatment. Molecular biomarkers represent alterations in gene sequences, expression levels, protein structure or function which can be used as to detect cancers at an early stage, determine prognosis, and monitor disease progression or therapeutic response.
Biomarkers which can predict the evolution of the disease and can help physicians decide which therapy is most likely to be effective for a given patient are invaluable tools for bothcancer research and clinical practice, yet few biomarkers are in clinical use despite decades ofintense effort (Tuma, 2008). Identification of appropriate biomarkers will enable evidence-baseddecision support for different treatment options considering risks, expected benefits and medicalcosts. Furthermore, clinical management of patients can greatly be improved, for example, for some tumors surgical removal is curative and adjuvant therapy is not necessarily, while for more malignant tumors aggressive systemic therapy, often chemotherapy, is required after tumor resection. The distinction between the more mild and the more aggressive BCa is not clear by usual clinico-pathological investigations, therefore decision support systems that can accurately predict the likelihood of recurrence and progression are urgently needed in the clinic.

MicroRNA as noninvasive biomarkers – MicroRNAs (miRNAs) are small, 19- to 23-nucleotidelong, single-stranded RNA molecules. It has been shown that miRNAs can affect the stability of messenger RNA (mRNA) and in some cases influence protein synthesis through partial sequence complementation with their interacting mRNA targets.. The specific regulation and biological function of miRNAs is largely unknown. The network of interactions is probablyas complicated as other biomolecular networks in the body. Unlike most classes of biomolecules, however there are far fewer known miRNA species. Like mRNAs, some miRNAs also show restricted tissue distribution; for example, miR-122 is highly enriched in liver, whereas miR-124 is preferentially expressed in neurological tissues. It has been shown that changes in the spectrum of cellular miRNAs correlate with various physiopathological conditions, including differentiation, inflammation, diabetes, and several types of cancers.
Currently, there are more than 1000 known miRNAs for humans (miRBase release 13.0), and they do not have known postprocessing modifications, which makes their composition much less complex than that of other biomolecules. Recently, some of the miRNAs previously identified in cells and tissues have also been found in extracellular fluids such as plasma, serum, saliva, and urine. The level and composition of these extracellular miRNAs again show changes that correlate well with diseases or injurious conditions. These observations suggest that extracellular miRNAs can be used as informative biomarkers to assess and monitor the body’s physiopathological status The ideal biomarker must be accessible using noninvasive protocols, inexpensive to quantify, specific to the disease of interest, translatable from model systems to humans, and a reliable early indication of disease before clinical symptoms appear.
Lower complexity, no known postprocessing modifications, simple detection and amplification methods, tissue-restricted expression profiles, and sequence conservation between humans and model organisms make extracellular miRNAs ideal candidates for noninvasive biomarkers to reflect and study various physiopathological conditions in the body.Therefore, miRNA profiles in serum and plasma samples from cancer patients have been screened to identify novel biomarkers for the diagnosis and evolution of tumors.

Modeling workflow – The main strategy of IntelUro consists in combining advanced artificial intelligence methods, in a knowledge discovery in data approach, with both mature genomics techniques like mRNA microarray and novel and promising methods like microRNA (miRNA) array. In order to attain its objective, the project uses an interdisciplinary approach, which requires a complex team including clinicians, pathologists, imaging doctor, molecular biologists and bioinformaticians. The IntelUro framework consists of an integrated solution for the following electronic medical recording software modules: clinical, imagining, pathological, and molecular biology catalyzed by a bioinformatics module.
The pathological, clinical, imaging and molecular module provide data that will be used by IntelUro to build classifiers for predicting progression and recurrence of bladder cancer. The performance of these classifiers strongly depends on the quality of data.

Bioinformatics module – High-throughput experiments investigate thousands of molecules in parallel. Bioinformatics tools must be used to select and rank a subset of molecules, hundreds or preferably tens, capable to discriminate between two or more medical situations. A challenging biomedical informatics problem is to use artificial intelligence to transform, interesting but not very useful, lists of ranked molecules into Intelligent Clinical Decision Support Systems. These should take as inputs the most relevant subset of molecules, and predict with the highest possible accuracy important clinical outcomes. In the following we enumerate and briefly describe the workflow associated with the bioinformatics module. It is based on the following steps.

– Data Collection and Integration.
– Data preprocessing
– Identifying differentially expressed genes.
– Clustering of genes and sample
– Annotation and placing differentially expressed genes on pathways/networks.
– Developing classifiers and intelligent clinical decision support systems

Leave a Reply

Your email address will not be published. Required fields are marked *