Early liquid biopsy in advanced non-small cell lung cancer: insights from the LIBELULE trial

The phase III LIBELULE trial assessed the clinical impact of implementing early liquid biopsy (LB) in patients with suspected advanced non-small cell lung cancer (NSCLC). While the study did not meet its primary endpoint—reducing time-to-treatment initiation (TTI) in the intention-to-treat population—it provided critical insights into the role of LB in precision oncology. In particular, early LB was associated with significant benefits in subgroups of patients, such as those with non-squamous histology, those requiring systemic therapy, and those harboring actionable genomic alterations (1).

In advanced NSCLC, molecular profiling is essential to determine eligibility for targeted therapies, which are associated with superior outcomes compared to chemotherapy or immunotherapy in patients with mutations such as EGFR, ALK, and ROS1. Tissue biopsy remains the standard for genomic testing but is often limited by procedural delays, sample insufficiency, or inaccessible tumor locations LB, which analyzes circulating tumor DNA (ctDNA) in plasma, offers a minimally invasive alternative that may accelerate diagnosis and support faster clinical decision-making. Technically, LB in this setting is supported by targeted next-generation sequencing (NGS) panels and digital polymerase chain reaction (PCR) assays that offer high analytical sensitivity for single-nucleotide variants, indels, and select fusions; emerging panels also quantify tumor fraction and report quality metrics that inform result interpretability (2-5).

Key advantages of LB include minimal invasiveness, rapid turnaround, avoidance of biopsy-related risks, and complementary yield when tissue is insufficient or inaccessible; importantly, early LB can help avert inappropriate first-line choices when actionable drivers are present (6,7).

The LIBELULE trial enrolled 319 patients with suspected advanced NSCLC. Participants were randomized 1:1 to receive either standard diagnostic workup alone or standard diagnostics plus LB at the initial diagnostic visit, using the InVisionFirst-Lung assay (a 37-gene NGS panel). The primary endpoint was TTI, defined as the time from randomization to treatment initiation based on molecular or pathological results. Secondary endpoints included detection of actionable genomic alterations, time to genomic results, and the impact of LB on treatment selection (1).

Because LIBELULE was conducted in France, its operational timelines and access to subspeciality care and targeted agents may not generalize to settings with different referral pathways, reimbursement frameworks, or molecular testing capacity.

In the intention-to-treat analysis, the primary endpoint was not met. Median TTI was 29 days in the LB arm vs. 34 days in the control arm (P=0.26). However, prespecified subgroup analyses revealed significant reductions in TTI. Among patients receiving systemic therapy, TTI was reduced by 9.8 days (P=0.011); in those with non-squamous NSCLC, by 10.8 days (P=0.010); and in patients with actionable alterations, by 16.6 days (P=0.003). Moreover, time to genomic results was significantly shorter with LB than with tissue testing (median 17.9 vs. 25.6 days; P<0.001), highlighting the potential of LB to expedite critical diagnostic information (8,9).

In the intention-to-treat analysis, the absence of an overall TTI difference likely reflects the inclusion of patients who ultimately did not require systemic therapy for lung cancer; among patients with confirmed NSCLC and those who received systemic treatment, early LB shortened meaningfully (10,11).

LB demonstrated a detection rate of ctDNA in 81% of patients. When integrated with tissue-based testing, actionable mutations were identified in 42.7% of patients with non-squamous NSCLC, compared to 36.5% with tissue testing alone. Notably, 13.6% of actionable alterations were identified exclusively through LB, demonstrating its value as a complementary diagnostic tool. Importantly, early LB also helped prevent inappropriate first-line treatment. In the control arm, two patients received immunotherapy despite harboring EGFR mutations—an error that likely could have been avoided with earlier molecular results from LB.

Despite these promising results, the trial also exposed several real-world challenges that hinder the full integration of LB into routine clinical practice. A substantial number of clinicians were hesitant to initiate treatment based solely on LB results, preferring to wait for tissue confirmation. This reluctance diluted the time-saving benefits of LB and emphasizes the need for greater clinician education and confidence in ctDNA assays. Moreover, LB is not without biological limitations. False negatives may occur, particularly in patients with tumors that shed low levels of ctDNA, such as those confined to the thorax or central nervous system. Consistent with LIBELULE, we advocate an upfront plasma-first approach for molecular profiling, reserving tumor NGS for cases with negative/indeterminate ctDNA or when orthogonal confirmation is clinically required (12-16).

From a logistical perspective, successful implementation of LB requires streamlined coordination between laboratories and clinicians, efficient sample processing, and rapid communication of results. Operational delays or uncertainty in result interpretation can undermine the potential clinical benefits of early LB, even when assays are technically successful. In addition, as the technology continues to mature, there is growing interest in expanding the capabilities of LB beyond the detection of single nucleotide variants and driver mutations. Emerging biomarkers such as gene fusions, tumor mutational burden (TMB), microsatellite instability (MSI), and minimal residual disease (MRD) may broaden the scope of LB and enable its use across multiple stages of cancer care—from diagnosis to monitoring, response evaluation, and early relapse detection (17,18).

While less commonly emphasized than false negatives, false-positive calls can arise from sequencing artifacts or clonal hematopoiesis; orthogonal confirmation and clinical correlation remain essential—particularly for unexpected variants at low allele fractions.

Endobronchial ultrasound-guided fine needle aspiration (EBUS-B-FNA) remains a valuable complementary procedure, enabling diagnosis/staging and provision of cellular material for confirmatory DNA/RNA testing when plasma is uninformative.

We briefly summarize ongoing ctDNA-guided trials in advanced NSCLC evaluating plasma-first selection for targeted therapy, escalation/de-escalation strategies and MRD-guided intervention; these studies are expected to delineate clinical scenarios where plasma guidance most improves outcomes.

When discussing NGS, we compare quality metrics between ctDNA and formalin-fixed paraffin-embedded/FNA (FFPE/FNA) (e.g., read depth, unique coverage, on-target rate, limit of detection and failure rates), to contextualize assay performance and decision-making (19,20).

Although a formal cost-effectiveness analysis specific to LIBELULE is pending, prior studies have indicated that upfront plasma genotyping may be economically favorable in selected patient populations, particularly when it prevents inappropriate therapies or reduces the need for repeat biopsies. Current clinical guidelines from the European Society for Medical Oncology (ESMO) and the National Comprehensive Cancer Network (NCCN) already support the use of LB in patients with advanced NSCLC when tissue is unavailable or when delays in tissue results may compromise care. Furthermore, both organizations advocate for broad-panel NGS—particularly in patients with non-squamous histology—as a standard component of molecular testing (7,12,21,22).

The LIBELULE trial therefore contributes important real-world evidence supporting the integration of LB into the diagnostic workflow for advanced NSCLC. While the trial did not achieve a statistically significant reduction in TTI across the entire population, its findings underscore the clinical value of early LB in specific subgroups where rapid molecular profiling can meaningfully impact treatment decisions. Avoiding ineffective or inappropriate therapies, particularly immunotherapy in patients with EGFR mutations, not only improves outcomes but also reduces patient risk and conserves healthcare resources (23,24).

Ultimately, LB should be viewed as a key component of modern precision oncology—one that enhances the timeliness, accuracy, and inclusiveness of molecular diagnostics. For LB to achieve its full potential in clinical practice, continued efforts are needed to address practical barriers, foster clinician trust, and ensure that diagnostic insights are translated into effective, personalized treatment strategies without unnecessary delays. As LB technologies continue to evolve and expand their biomarker repertoire, their role in the personalized management of advanced NSCLC is expected to grow (25).

An important next step is the deployment of technical and pre-analytical solutions that reduce false-negative ctDNA results [e.g., deeper unique molecular identifier (UMI)-based sequencing, optimized tube handling, tumor-fraction estimation], which will further curtail the need for tumor NGS, accelerate TTI, and lower overall testing costs.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.