LIBELULE: A randomized phase III study to evaluate the clinical relevance of early liquid biopsy (LB) in patients with suspicious metastatic lung cancer

9019 Background: Timeliness of molecular testing is essential to minimize the time to appropriate 1 st line treatment initiation (TTI) in advanced NSCLC pts. We hypothesized that LB-based molecular testing performed at the patient's 1 st visit could reduce this TTI. Methods: LIBELULE is a multicenter, randomized, comparative, open-label study, enrolling pts with radiological suspicion of stage IV lung cancer, no prior biopsy or cytology for advanced NSCLC diagnosis. In Arm A, a LB was performed at the 1 st visit using the InVisionFirst-Lung assay, an amplicon-based NGS panel covering 37-NSCLC associated genes. When LB results were informative, treatment was initiated on the basis of: InVisionFirst-Lung alone for pts with EGFR, BRAF V600E mutation (mut), ALK or ROS1 rearrangement (category 1 alterations); InVisionFirst-Lung and pathological results for pts with ERBB2, MET exon 14, KRAS, BRAF non V600 and/or LKB1 mut, RET or NTRK rearrangement (category 2 alterations). In control Arm B, histological sampling was planned with genomic analysis when indicated (local LB allowed) and treatment was initiated according to ESMO guidelines. The primary endpoint was the time from randomization to initiation of appropriate treatment based on informative genomic and pathological results. 286 pts were needed to detect a 21% shortening in TTI (α = 0.05, 1-β = 90%). Results: 319 pts were randomized between Arm A (n = 161) and B (n = 158): median age 68 y (39-97), 56.1% male, 28.5% non-smokers, PS0-1 82%, PS≥2 18,1%; adenocarcinoma: 56.7%, squamous cell carcinoma: 11%, SCLC: 10%, other tumor types: 5%. 5.3% of pts had no cancer. In Arm A, 81% of pts had ctDNA findings; category 1 and category 2 alterations were identified on tissue and/or LB in 29.2% ( EGFR mut: 21.7%) and 24%, respectively, whereas in Arm B, 23.2% of pts had category 1 ( EGFR mut: 20.3%) and 20% had category 2 alterations detected. Systemic treatment was initiated in 74.5% and 65.8% of pts in Arm A and B, respectively. Main reasons for not initiating treatment were diagnosis other than cancer, local treatment and palliative care. The mean TTI was 29.0 days (d) (95%CI 25.9-32.1) in Arm A vs 33.9 d (95%CI 28.4-39.5) in Arm B in the intention-to-treat population (p = 0.28), but significantly shorter in Arm A 29.1 d vs 38.8 d in Arm B for pts receiving systemic treatment (p = 0.01). The mean TTI was significantly shorter (21 d vs 37.4 d) in pts with category 1 alterations (p = 0.004) as well as the time to contributive genomic analysis (17.9 d vs 25.6 d; p < 0.001) in Arm A and B, respectively. In Arm A, 7.4% of pts vs 13.3% in Arm B initiated a treatment without genomic analysis available. Conclusions: Early LB significantly reduces the time to contributive molecular analysis and the time to initiation of appropriate 1 st line therapy in pts eligible for systemic treatment, especially for pts with actionable alterations indicating targeted 1 st line therapy. Clinical trial information: NCT03721120 .