Incremental clinical yield of serial large NGS panel liquid biopsies in routine care: Analysis of the STING study.

3046 Background: Plasma ctDNA profiling is increasingly repeated over the disease course to capture tumor evolution and newly targetable alterations. However, the incremental clinical yield of serial ctDNA profiling in routine practice remains incompletely characterized. Methods: This analysis was conducted within the prospective STING study (ClinicalTrials.gov identifier: NCT04932525), a program that aims to identify actionable molecular targets in patients with advanced solid tumors. Eligible patients were adults with advanced solid tumors and provided written informed consent for molecular analyses. Circulating tumor DNA (ctDNA) profiling was performed using an FDA-approved hybrid-capture next-generation sequencing assay, enabling to analyze over 300 genes, and assessment of tumor mutational burden (TMB) and microsatellite instability (MSI) from plasma samples. All molecular results were reviewed during a weekly institutional Molecular Tumor Board to assess clinical relevance and therapeutic implications. For the present study, patients with ≥2 FMI liquid biopsies were included. The earliest ctDNA test was defined as baseline (BL1), and the second earliest as BL2. Actionable alterations were classified according to the ESMO Scale for Clinical Actionability of Molecular Targets (ESCAT). The primary objective was to determine the proportion of patients in whom repeat ctDNA testing identified new actionable alterations not detected at BL1, and their ESCAT tier distribution. Results: Among 1,565 patients (3,494 liquid biopsies), the median number of tests per patient was 2 (range 2–6). The most common tumors were lung (26.4%), breast (13.1%), prostate (10.9%), colorectal (CRC) (8.0%), pancreas (6.7%), and thyroid (4.1%). The median BL1→BL2 interval was 313.5 days (range 2–2,279). At BL1, 470 patients (30.0%) harbored ≥1 actionable alteration (ESCAT I–III); excluding TMB-high, 23.9%. Repeat testing identified ≥1 new actionable alteration in 316 patients (20.2%); excluding TMB-high, 10.7%. Notably, 227 patients (14.5%) converted from no actionable alteration at BL1 to ≥1 actionable alteration on later testing (8.1% excluding TMB-high). Newly detected actionable alterations were distributed as ESCAT II (54.3%), ESCAT I (26.1%), and ESCAT III (19.7%); 93 patients (5.9%) gained at least one new ESCAT I alteration. The yield of new actionable alterations varied by tumor type (~29% CRC, 25.9% breast, 23.5% lung, 14.6% prostate). Conclusions: In a large real-world cohort, serial ctDNA testing provided clinically meaningful incremental detection of actionable alterations in ~1 in 5 patients, including new ESCAT I findings in ~6%. These data support serial ctDNA profiling to capture evolving tumor genomics and expand therapeutic opportunities. Prospective studies should define optimal timing and confirm downstream clinical impact.