Oncology Translational Laboratories Molecular Pathology Information Management System

Oncology Translational Laboratories: Molecular Pathology Information Management System © Raj Bandaru 2010 View of Boston skyline from Novartis Raj Bandaru, Sonia Agrawal, Humphrey Gardner MD. OTL Statistics and Data management, Novartis Institute for Biomedical Research. San Diego, Pathology Visions 2010. Oct. 25.

Biomarkers of increasing significance Novartis commitment to biomarkers in Treating patients • Biomarkers in drug discovery and therapeutic paradigms: High priority effort at Novartis. • Numerous clinical trials underway focused on getting the best treatments to the right patients. • Deeply invested in measuring and understanding various biomarkers in all our clinical studies (Pre, Phases 1 – 3 and post marketing). • A highly data (biomarker) driven process that involves sampling and collection of a large volume of data driving real time decisions. • The Oncology Translational labs: center of excellence in Novartis to ensure high quality and timely data delivery for making critical treatment decisions • Large part of this data is generated within our own labs to ensure high quality and fidelity across studies. 2 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Customer: Oncology Translational Labs Multi site / multi platform integration. . . In real time • Tissue Biomarker Group (IHC), • Molecular Pathology Lab, (CAP) • Molecular Profiling Lab, (CAP/CLIA) • DNA analyses/mutation detection Cambridge, MA East Hanover, NJ • ELISA and proteomics Tissue Biomarker Group • Flowcytometry Lab (IHC) • Gene expression Profiling Basel, Switzerland 6 pathologists, 12 Histotechs. . Over 60 users across 3 sites 4 Aperio ® scanners 9 Ventanas ®, 2 Nemesis 2 CRI ® scanners/ imaging stations 2 3 D Histech ® scanners, Laser Microdissection Circulating Tumor Cells, Veridex Celltrax ® Sequenom ®, Fluidigm ®, Dxs, RTPCR, Affymetrix ®, Nanostring® ABI ® sequencers. . . Powerpath ® (validated database for CAP lab) ~ 15000 samples per year (~ 70, 000 slides) Validated assays for > 30 analytes, FISH, CISH, mutation/copy number analyses = Millions of data points! In hundreds of Spreadsheets, emails, and other documents 3 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Life before the LIMS Spreadsheets, emails. . . Perl scripts • Samples and data recorded and tracked in Excel Spreadsheets. • Laboratory processes not tracked or in systems not accessible (Ventana system) • Pseudo audit trails (periodic time stamped back up of all data files using local scripts) • Multiple data entry, sample tracking sheets, Ventana staining systems, Aperio image database and Pathologist data entry. • Data transfer and sharing via emailing of spreadsheets • Data uploads to validated systems involved multiple independent and repeated cross checks using non validated scripts. 4 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Tough Life Errors, duplication of effort, redundancy and confusion • Limited sharing/ security/audit trails/back up of the data • Structure of folder / files not defined by the process in the labs • Non transferable/ searchable data (color coding/ formatting of cells) • No controlled vocabulary/ validation of data entries. • Identified several errors and missing data (not entered or accidentally deleted) in 2 years of legacy data. (from spreadsheets prior to 2006). • Very difficult to implement automation, triggers, reminders. © Raj Bandaru 2010 “Black clouds over San Diego Bay” taken with Droid. X, Oct. 23 rd AM 5 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Off the shelf solutions Gaps IHC. . . FISH. . . Copy number. . . RNA profiling. . DNA sequencing. . . Range of Biomarker Assays Primary limitation of off the shelf systems was the need for extensive customization, expensive maintenance contracts/ consultation fees. . And still require internal support teams. Watson LIMS Molecular Profiling Assays Histo Pathology Assays Powerpath Aperio Spectrum Ventana LIMS Increasing Automation/Digitization 6 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only What was missing

Phased implementation step by step development with the big picture in mind. Data sources/ Labs IHC Aperio ® STS Enrollment Log OTL db SMU RNA Clinical Statistics data export Spotfire® Flow ELISA Independent database Imaging Misc. spreadsheets External data Preclinical/ cell line data etc. 7 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only Reports Analysis Datamining R toolbox

Key Requirements What Information is tracked • Log and track samples, child samples, arrays and sample meta data (subject information, diagnosis), by studies/ projects, tissue microarrays. • Create tasks/experiments, request tasks, log task metadata and process details. • Two way communication link between Lab instruments and the LIMS. • Store results/ data features associated with samples and compiled data sets and associated analyses programs (SAS ® / R ® / Inform ®) and reports with e-signatures • Link to external sample related data (images, pathology reports). • Link to data mining/ analyses/visualization engines (Spotfire ®) • Multilevel security, access restrictions at data and functionality level. • Audit trail of edits, frequent back ups of data with recovery plans. • Flexible and complex query interface to generate reports. • Multisite synchronization of data, load distribution and performance opt. • Automated emailing/ reminders / task tracking/ report generation. • Customizable control vocabulary, data features, form display formatting. 8 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Interim Solutions Preparing the labs for change • Implemented automated backups and versioning of the spreadsheets and data files. • Updated spreadsheets automatically with data entered to the database. • Partial integration with Aperio ® Spectrum via barcode scanning and scripts to synchronize databases. • Implemented Powerpath ® a validated database for the CAP/ CLIA certification. • Some limited customization of Spectrum ® and Powerpath ®. • Set up scripts to extract data from Ventana ® run reports and load into the LIMS. • Started synchronizing sample identifiers across labs. 9 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only © Raj Bandaru 2010

Architecture Validation focussed • Windows server running Apache 2. 0 Web server and PHP • 64 bit Windows server running MSSQL 2005. (dev and prod instances) • Plans to migrate production systems to validated servers (2011) • Low level PHP/javascript/AJAX : highly structured code. • Interface design, structure, actions, validation rules and function calls stored within tables in database • High on performance and ease of use, Low on bells and whistles. • Highly modular, easily maintainable code. (Basic PHP developer friendly), Code folders structure reflecting interface design, all SQL in one location/class. Globals functions in specific classes Database Lookup form elements, actions, form layout specified in table form generator form processing 10 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only form editing

Design CFR 21 Part 11 • Approximately 65 tables and ~ 10 views plus audit trail. • No records are ever deleted from the database, just moved to mirrored set of tables. • Access restrictions by user/ user groups at the level of tables, fields in a table or specific data elements. • Very minimum server dependent functionality to enable easy porting to other database servers. • Normalized database with Optimized queries and compound indexing. 11 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Interface Simple and low learning curve • Followed an Excel spreadsheet like feel, but via the web interface. • Minimal user key strokes for data entry. • Arrange tabs in order of workflow in the labs. • Highly flexible query interface capable of complex searches with unlimited filters and user defined reports. • Ability to save searches and set up cron jobs to run searches/reports. • User customizable web interface, formatting of fields, yet simple look and feel. 13 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Timelines and Resources May 2007 May cti on Re IH lea C se sa V m er ple sio tra n 1 ck : ing Re Ta l. V sk er M. 2 an ag em Re en Ve l. V t nt er. an a 2. 1 In te gr at ion Sept. 2008 May Oct. 2009 April 2010 DN A lab un ut f inp ta Da ta Be St ar te d Re De lea se ve lop m en t Rapid development/ testing cycles Jan. 2011 • On average one or two full time developers at any point. Indicated by blue arrows above. • Long term plan is to make code open source community include in a public library for everything from LIMS to image and data analyses for the Digital Pathology community 14 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

Light at the end of the tunnel Happy Pathologists and Histotechs • Database contains over 120, 000 records. (All legacy data ported over). • Modular code has allowed for rapid development of new functionality (< 1 week from design to testing and production release) • Far fewer errors or mistakes in the labs. Zero data errors. • Some database performance issues to be resolved. • Integration of other laboratory instruments and imaging stations. © Raj Bandaru 2010 15 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

What is Ahead Automation everywhere. . . • Complete CFR 21 part 11 Validation of the entire system. • Reporting (Spotfire/business objects) and extend interface and reporting on mobile devices • Web access to analyses scripts and tools (SAS, R, Matlab code). • Implement processing tasks creation logic interface. • Wider acceptance of barcoding/ standardization of procedures. • Text mining and knowledge management tools. Integrate with other imaging and biomarker databases. . . MICAD, protein atlas, ncbi GEO, Pubmed. • General cosmetic/performance improvements. Google like searches. © Raj Bandaru 2010 16 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only

What is Ahead Integration across technologies Image Analyses with DNA analyses processes. Track entire flow and process in the LIMS Automated Tumor finding algorithms. (CRI ® inform) Pathologist verify LCM and microdissect tumor region Histotech supervise Send for DNA analyses Analyses of Circulating Tumor cells and accessibility of images and results to Pathologists. HSP 70 CK-PE CD 45 APC Processing of TMA core and Fluorescence images (CRI ® software) Track and store application settings, data outputs and resulting images. Make results available via visualization tools. 17 | Oncology Translational Labs IMS| Raj Bandaru | 20101024 | Informatics | Business Use Only