transcend: the role of informatics in complex clinical trials

TRANSCEND: The Role of Informatics in Complex Clinical Trials:

TRANSCEND: The Role of Informatics in Complex Clinical Trials The TRANSCEND Team

I-SPY 2 Protocol – The genesis of TRANSCEND:

I-SPY 2 Protocol – The genesis of TRANSCEND

I-SPY 2 in a nutshell:

I-SPY 2 in a nutshell I-SPY2 – a breast cancer treatment trial Goals: Accelerate knowledge discovery remove the need to submit multiple protocols that only vary by agent type. create a ‘pipeline’ model where changing agents does not require a new protocol Integrate biospecimen management into the trial infrastructure manage specimens starting at the point of collection through the lab and resulting process Accelerate knowledge dissemination Allow access to data as the trial progresses – scientist portal with data ‘published’ to the portal as patients are “post novel therapy”. Provide data with standard encoding – over 600 data elements mapped to caDSR (and Snomed CT) provide the data through a simple process that adheres to the trial’s data sharing policy – a web-based secure scientist portal

What was needed...:

What was needed... I-SPY 2 Clinical Trial required the following: A system that provided controlled data entry “up front” to avoid “retrospective information foraging” A system enforcing complete enrollment/eligibility data prior to randomization An automated randomization ‘engine’ Extensive tracking of the participants - knowing the “state” of each participant so delays in standard therapy are avoided Integrate specimen management Data mapped to standard representations (standardized data elements) so data can be merged across other similar study data sets (ie, use standard coded representation for weight, height, diagnosis, etc..) Structured data entry of adverse events in order to provide computable and complete adverse event data using CTCAE v4 make the research data set available to other scientists in a far shorter time than is typical (available before publication and in a useful form through a simple process) 4

What is TRANSCEND?:

What is TRANSCEND? TRAN slational informatics S ystem to C oordinate E merging biomarkers, N ovel agents, and clinical D ata Qualifying & Exploratory Biomarkers

Novel Informatics Aspects of TRANSCEND :

Demonstrate an information management infrastructure that supports complex clinical trials like I-SPY 2 Experiment with a “clinical entry mode” where clinicians enter data in the course of clinical documentation and CRFs are automatically populated removes the ‘data abstraction’ step merges ‘clinical documentation’ with ‘clinical trial documentation’ Leverage existing NCI (caBIG) freely available software systems as part of an integrated ‘system’ in support of a large trial Demonstrate the successful use of an “automated randomization engine” in a clinical trial Novel Informatics Aspects of TRANSCEND

TRANSCEND ‘tech facts’:

TRANSCEND ‘tech facts’ It is a single “tech stack” housed in a data center in California and used by all institutions (no local deployment – only requires a web browser for users) – think “Software as a Service (SaaS)” model... It was built on time and on budget (deployed July 2010) first I-SPY 2 patient enrolled March 2010 (UCSF) Four institutions using the system, two more going live this month Nearly all components are open-source software (all except the randomization engine) Used the following open source systems: caBIG tools: caTissue, caHub (caXchange), caIntegrator, caARRAY) Tolven eCHR: an open-source, configurable EMR Minimal “programming” – mostly assembly, mapping to caDSR elements, data capture template design, establishing workflow 300 hours of programmer time (7 weeks) ~1,500 hours of design time (requirements gathering, architecting, assembly, mapping, form configuration) 7

Slide 8:

TRANSCEND Summary Screen

Slide 9:

TRANSCEND Case Report Form Selection Menu

Clinical Documentation Templates:

Clinical Documentation Templates

Randomization control:

Randomization control

Randomization by remote service:

Randomization by remote service

Documenting AE’s with common terminology- CTCAE (v4):

Documenting AE’s with common terminology- CTCAE (v4)

Tissue Specimen Form:

Tissue Specimen Form

Specimen Tracking:

Specimen Tracking

Slide 22:

caTissue – Managing Biospecimens in the lab

Making Clinical Trial Data Accessible (caINTEGRATOR):

Making Clinical Trial Data Accessible (caINTEGRATOR)

Lessons Learned:

Lessons Learned Clinical Trial “Informatics” needs to be viewed from an enterprise-level perspective which stresses integration across multiple components (trial registration system, biospecimen management system, EHR, research portal/data set library system) caBIG tools were designed as ‘mission specific tools’ that could be put on a grid, but were not originally designed interoperate to allow completion a workflow task that required some or all of them together. Adaptive trial designs require automated randomization which need substantial “testing” to ensure validity/robustness we perform over 100 regression tests on the randomization engine per release CRF’s can be populated from *correctly constructed structured* “clinical documentation” templates used by clinicians

My Wishlist – “plugging into instutional infrastructure”:

My Wishlist – “plugging into instutional infrastructure” EMR Integration at each site limited by prevailing HL-7 infrastructure “on the ground” (mostly v2.x) that does not have messages designed for this purpose limited by funding/time – 20+ institutions = building and managing 20+ bi-directional interfaces limited by structured data capture capabilities in prevailing EHRs (or institutional configurations or strategy that limits forced structured data capture to facilitate adoption by clinicians used to narrative text documentation) Biospecimen management system integration at each site limited by variable adoption of these systems across the participating systems limited by a lack of standards around interoperability between BioMgmt Systems and EMRs. Integration with institutional adverse event reporting infrastructure limited by high variability in this process (manual to fully electronic) across the institutions caARES lacked a web services interface when we built TRANSCEND

Acknowledgments:

Acknowledgments Meg Young – TRANSCEND Project Manager (UCSF) Kathy Hajopoulos – Project Oversight (UCSF) Sarah Davis – I-SPY 2 UCSF Trial Manager (UCSF) Sarah Boortz – Testing (UCSF) Mike Morris – Design/Requirements (UCSF) Sorena Nadaf – Informatics/Design (UCSF) Dr. Angie DeMichelle – Clinical Oncology (U Penn) Kyle Walthen – Randomization Engine (MD Anderson) Ashwin Koleth – Software Development (Tolven) Dan Milgram – Trial oversight (CCSC) John Koisch – Architecture (NCI) Christos Andonyadis – Architecture (NCI) NCI CAT Committee members - 2009 Nancy Roche – Project Oversight (SAIC) Dr. Laura Esserman – Principal Investigator (UCSF) In recognition of its novel approach and design principles, TRANSCEND was awarded the National Cancer Institute's 2010 caBIG "Innovation Award". TRANSCEND funded by the National Cancer Institute (NCI), Subcontract # 28XS197