P2.16-05 Population-Based ROS1 Testing in Lung Cancer: Creating Opportunity in a Publicly Funded System

Abstract

ROS1 gene rearrangement is found in 1-2% of all non-small cell lung cancer (NSCLC) and is recommended as standard molecular diagnostic testing. This study models the most efficient ROS1 testing strategy to maximize detection of true positive cases (TP) while minimizing costs and turnaround time (TAT). A population-based ROS1 testing model was developed from a Canadian (Ontario) public healthcare system perspective examining the use of immunohistochemistry (IHC) and next generation sequencing (NGS) versus fluorescence in situ hybridization (FISH, gold standard), reflex versus molecular or clinical selection (never smokers), and blood- versus tissue-based testing. Model inputs were derived from existing literature and expert opinion. Direct testing costs and TAT were obtained from the Ontario public perspective (University Health Network, Cancer Care Ontario). The most cost-effective strategy for the outcomes of TAT and TP was reflex testing with IHC and subsequent FISH confirmation; this identified a high proportion of TP within a relatively short TAT. The most costly reflex strategy was NGS, with a greater proportion of missed TP (Table), and long TAT. Clinician selection of never smokers yielded the lowest proportion of TP. Population-based plasma ctDNA testing using commercial assays was the most costly strategy. One-way sensitivity analysis demonstrated that the TP outcome was most sensitive to the population prevalence of ROS1, while cost was most sensitive to the specificity of ROS1 IHC. Pathologist-initiated reflex testing using IHC with FISH confirmation provides the most cost-effective population-based testing strategy. Clinician-initiated testing significantly lengthens TAT. Selecting only never smokers for testing misses a larger proportion of TP patients who would benefit from targeted therapy despite potential cost savings.