Real-world Clinical Effectiveness of Whole Genome and Transcriptome Analysis to Guide Advanced Cancer Care

Overview

This study aims to determine the clinical effectiveness of whole-genome and transcriptome analysis (WGTA) to guide advanced cancer care. The study setting is the British Columbia (BC) Personalized OncoGenomics (POG) program, a single group research study of WGTA guiding treatment planning for patients with advanced, incurable cancers (NCT02155621). To characterize clinical effectiveness, the survival impacts of POG's approach compared to usual care in matched controls will be estimated.

WGTA provides an opportunity to improve health outcomes for patients by tailoring treatments to each individual's genomic profile. The BC POG Program is a single arm research study integrating WGTA information into clinical decision-making for patients with advanced stage, incurable cancers.The clinical effectiveness of POG's approach is unknown. This retrospective quasi-experimental observational study will estimate the real-world effectiveness of WGTA for guiding advanced cancer care. To identify a counterfactual for POG's single-arm approach, matching methods combined with administrative healthcare data will be used. The survival impacts of POG's approach compared to usual care in matched controls will then be estimated. Specific Aims and Hypotheses This study aims to estimate the overall survival effects of POG's approach versus usual care for patients with advanced cancers. Hypothesis (null): there is no difference in survival across POG and usual care patients Hypothesis (alternative): POG patients who initiated WGTA live longer, on average, than usual care patients Study Design This study will apply a retrospective cohort design. Cohorts will include patients who consented to POG and underwent a biopsy for WGTA between July 2014 and December 2017 and matched usual care controls. POG patients who enrolled prior to July 2014 will be excluded from our study because during this feasibility period, referring clinicians employed a high level of case-by-case recruitment selection. Usual care patients will be matched to POG patients using supervised learning techniques. The study period will range from patient's time of metastatic cancer diagnosis to December 31 2018. Data Sources De-identified linked population-based administrative datasets will be obtained from BC Cancer for all adult patients (\>18 years) diagnosed with cancers in BC prior to December 2017. POG patients will be identified from the BC Cancer Outcomes and Surveillance Integration System (OaSIS) POG Module Database. Eligible control patients will be identified from the BC Cancer Registry, a population-based provincial cancer registry. These data will be linked with data from the BC Cancer Pharmacy Database, Radiotherapy Database, and Cancer Agency Information System (CAIS) using agency-specific identifiers. Statistical Approach The investigators will match POG patients and usual care patients based on their date of metastatic disease diagnosis. They will apply 1:1 genetic algorithm-based matching (1:2 in sensitivity analysis) and match patients on propensity scores and baseline covariates, including patient demographics, clinical characteristics, treatment histories, and healthcare utilization prior to metastatic disease diagnosis. When necessary, matching analyses will be stratified to account for variation across cancer types. To estimate overall survival in POG patients and matched controls, non-parametric and parametric survival analyses will be used. These analyses will be adjusted for censoring. The investigators will explore heterogeneity in clinical effectiveness across cancer subtypes through subgroup analysis and use scenario analysis to determine the impact of future changes in the application of WGTA on clinical effectiveness.