Pan Zhao

Bioinformatics, Vishuo Biomedical Pte Ltd.

Objective:Cancer is one of the leading causes of mortality and morbidity worldwide. Precision cancer therapies are expected to increase the efficacy and safety of a drug when given the specific biomarker information of a patient. Whether the patient will respond to or resist a given targeted therapy can largely be predicted. Currently, no one integrated database that contains disease specific biomarkers, accurate clinical annotation of the biomarkers, targeted therapy drugs against these biomarkers and has ability to build a local knowledge base is available. We aimed to build such an integrated database with a user-friendly interface and ability to build a knowledge base locally. The system will aid clinicians in treatment decision-making process and may also serve to guide future therapeutic approaches. Method: iCMDB contains six groups of data: (i) drug and disease biomarkers used in clinical studies: the information is collected from FDA Pharmacogenomics Biomarkers in Drug Labels and diagnostic biomarker prognostic biomarker database. (ii) Mutations with clear clinical evidence: the information collected from drug clinical trial result with large proportion from FDA database submitted by pharmaceutical companies, PUBMED articles for clinical report of using different drugs and NCI drug information. This information associated with drug and targeted biomarkers include drug stage, disease stage, drugs, PFS (progress free survival), OS (overall survival), compare group in clinical trial, patient numbers, hazard ratio etc. (iii) drugs approved by FDA or in clinical phase II and III: drug information is derived from NCI drug dictionary, drug bank, drug clinical trial phase II and III data and cancer research UK. The drug data released by pharmaceutical companies are also collected into the database. (iv) Cancer treatment plans: the information is found in NCCN guideline [http://www.nccn.org] or reported in medical journals: (v) patient genome sequencing analysis result: the build-in NGS data analysis pipeline can support both illumina and life tech platform. To provide precise analysis result, iCMDB system processes the analysis result by multiple filters. (vi) Patient therapy data uploaded by system users for reference and further study. Result: The platform was built and become functional in March 2012. Since then, over 10 hospitals have tested the system and found it to be robust and accurate. They have been using the system for routine patient reporting. Conclusion: We have shown that an integrated platform containing expert curated biomarker, clinical annotation, drug and treatment database helps greatly with cancer patient management. We spent 4 years building and improving such a system. We also build in the function of forming a local knowledge base in this system, so that hospitals and clinicians benefit from the increasing number of samples stored.

Key Words: big data  bioinformatics  data base