Clinical Trial Analytics

EACRIProvidence Cancer Center, PPMC (Portland, OR)

Summary

The purpose of the EACRI clinical analytics program is to transform the vast amounts of raw data generated by clinical trials into structured data sets, then mine those data sets for actionable insights that enhance trial efficiency, maximize therapeutic benefit and improve the odds of achieving regulatory approval.

Overview

The Providence Cancer Center partners with a wide range of sponsors, from large pharmaceutical companies to non-commercial entities, to conduct drug trials that leverage our clinical expertise and comprehensive portfolio of cutting-edge analytical technologies. While large pharmaceutical companies have proprietary systems for capturing, integrating and analyzing the data generated from their clinical trials, many smaller companies and non-commercial trial sponsors lack these capabilities. To address this need, our team has developed solutions that allow us to integrate virtually all of the data streams generated by our clinical trial program and those generated by external sources. We have also developed advanced data-mining capabilities that allow us to deliver the following insights:

  • Identification of specific cohorts of patients which are more likely to benefit from a given therapy.
  • Identification of specific cohorts of patients which are more likely to experience adverse effects (clinical and experiential).
  • Predictive biomarkers that identify which patients are responding positively to therapy and/or have an elevated risk of experiencing an adverse event.
  • Identification of possible combination therapies that have a high potential for synergy.
  • Identification of possible alterations to the dosing schedule that may improve efficacy and/or minimize adverse events.

Availability

Our data integration, database development and analytics capabilities are available to our pre-clinical and clinical trial partners here at the Providence Cancer Center. For more information, contact Dr. William Redmond (William.Redmond (at) providence.org). We are also creating free, open-source standalone versions of some of our analytics tools for use by the wider cancer research community. These public tools are designed to accept user-generated CSV (Excel) files, and are available @ http://www.mycancerproject.org/boss/

Compatible Data Sources

The development and application of cancer immunotherapy at the Providence Cancer Center leverages a number of powerful new technologies, from advanced immunological monitoring to next-generation genetic sequencing (NGS). These advanced analyses complement the traditional portfolio of clinical parameters involved in assessing the efficacy, safety and mechanisms of an investigational therapy. Our clinical analytics team can integrate all of the data from these in-house sources (and virtually any external data provided by the sponsor)into a unified database that can be mined for actionable insights. Our platform is capable of integrating data from the following sources:

“Omics” Data

We can normalize and integrate data from almost every type of “omics” data set, including: NGS/RNAseq, NanoString, mRNA/miRNA/SNP arrays, protein/antibody/cytokine arrays, proteomics, metabolomics, etc.

Electronic Medical Records (EMRs)

Results from standardized clinical assays (eg. complete blood counts, lipid panels, etc) and many types of patient history and demographic data can be integrated in bulk. Non-standardized data (eg. Physician notes) can also be integrated, but require manual standardization.

Flow Cytometry

Flow cytometry is perhaps the most powerful tool for immunological monitoring and the most refractory to standardization. We have developed a proprietary data pipeline that allows us to integrate flow results from different sources (using different antibody panels and equipment) into a single data set. Because gating strategies for flow cytometry data are highly subjective, our system allows for the submission of multiple interpretations of the same data set.

Outcomes

Any kind of outcome (eg. OS, DFS, ORs, irAEs, quality of life indicators, etc) can be integrated. Outcomes can be temporal (eg. adverse event at a specific timepoint) or non-temporal (eg. total number of adverse events recorded during treatment).

Histology

Conventional pathology reports and data sets from multi-spectral analyses (Vectra/inForm, etc).

Patient Experience

Conventional quality of life (QOL) survey data and customized patient survey platforms. Direct data capture from patient-facing electronic surveys.

Pre-Clinical Data

Data from mouse, primate and other animal trials.

Analytics and Outputs

In addition to delivering actionable insights, the goal of our analytics program is to provide stakeholders the direct ability to visualize and interrogate their data sets. To facilitate this, we have built a number of web-based tools that render interactive visualizations using information from the underlying database. The objective is to simplify this experience to the point that is readily accessible to stakeholders at different levels of scientific and computational literacy. Some of the key features of this system are:

  • Web-accessible. No special software, user interface runs in Google Chrome. Select analytics available on mobile devices.
  • Simple, interactive, visual interface. Built-in statistical analyses. Multiple ways to visualize data.
  • Analytics automatically update as new information is added to the database.
  • Integrated data sets can be exported for analysis with 3rd software (eg. SAS, Prism).

Visualizations

We have developed multiple methods to visualize, analyze and communicate data generated by clinical trials, and cancer research in general. For interactive demonstrations of many of these visualization tools, visit www.mycancerproject.org/visualizations/

Timecourse

Timecourse Feb8

Correlations and Relationships

Relationships Feb8

Cohort and Group Comparisons

BOSS Comparisons