Cell-Free DNA at Diagnosis for Stage IV Non-Small Cell Lung Cancer: Costs, Time to Diagnosis and Clinical Relevance

SIMPLE SUMMARY: To obtain non-small-cell lung cancer (NSCLC) tissue, patients undergo burdensome procedures, such as needle biopsies or endoscopic or surgical procedures. Recent technological developments have enabled targeted sequencing using plasma. Next-generation sequencing (NGS) based on cfDNA can provide broad genetic information and comprehensive identification of relevant targets for therapy in a single test. Consecutive or combined use of liquid biopsy-based NGS can potentially result in more efficient and less invasive molecular profiling in NSCLC patients. Using a process-based discrete event simulation with data on stage IV NSCLC patients from the LEMA trial, this study aimed to estimate the effects on costs, throughput time, and diagnostic yield of two diagnostic scenarios with liquid biopsies, compared to diagnostics with tissue biopsies alone. ABSTRACT: Tissue biopsies can be burdensome and are only effective in 10–30% of patients with metastasized non-small-cell lung cancer (mNSCLC). Next-generation sequencing (NGS) on cell-free DNA (cfDNA) might be an attractive alternative. We evaluated the costs, throughput time, and diagnostic yield of two diagnostic scenarios with tissue and cfDNA for mNSCLC patients, compared to diagnostics based on tissue biopsy alone. Data were retrieved from 209 stage IV NSCLC patients included in 10 hospitals in the Netherlands in the observational Lung cancer Early Molecular Assessment (LEMA) trial. Discrete event simulation was developed to compare three scenarios, using LEMA data as input where possible: (1) diagnostics with “tissue only”; (2) diagnostics with “cfDNA first”, and subsequent tissue biopsy if required (negative for EGFR, BRAF ALK, ROS1); (3) cfDNA if tissue biopsy failed (“tissue first”). Scenario- and probabilistic analyses were performed to quantify uncertainty. In scenario 1, 84% (Credibility Interval [CrI] 70–94%) of the cases had a clinically relevant test result, compared to 93% (CrI 86–98%) in scenario 2, and 93% (CrI 86–99%) in scenario 3. The mean throughput time was 20 days (CrI 17–23) pp in scenario 1, 9 days (CrI 7–11) in scenario 2, and 19 days (CrI 16–22) in scenario 3. Mean costs were €2304 pp (CrI €2067–2507) in scenario 1, compared to €3218 (CrI €3071–3396) for scenario 2, and €2448 (CrI €2382–2506) for scenario 3. Scenarios 2 and 3 led to a reduction in tissue biopsies of 16% and 9%, respectively. In this process-based simulation analysis, the implementation of cfDNA for patients with mNSCLC resulted in faster completion of molecular profiling with more identified targets, with marginal extra costs in scenario 3.