ALFA score as a predictor of long-term benefit with frontline brigatinib in ALK + advanced NSCLC: A ctDNA-based composite biomarker from ALTA-1L.

8646 Background: Optimal sequencing of ALK-TKI in 1st-line ALK+ NSCLC remains controversial. Although 3 rd -generation (G) TKI are increasingly used upfront, a subset of patients (pts) may achieve prolonged outcomes with 2 nd G TKI. However, baseline (B) biomarkers to identify these pts are lacking. We evaluated whether a composite clinical-molecular score integrating (B) ctDNA and biological parameters could predict systemic and intracranial outcomes with upfront brigatinib. Methods: Post hoc analysis of pts with ALK+ NSCLC from the ALTA-1L trial with available (B) ctDNA NGS. ALK variants, co-occurring genomic alterations (co-GAs), and clinical/biological variables (PS, albumin, derived neutrophil-to-lymphocyte ratio [dNLR], and metastatic sites) were analyzed. Independent prognostic factors for progression-free survival (PFS) and intracranial PFS (IC-PFS) were identified by multivariable Cox models. A composite score (ALFA), based on the Cox models’ coefficients was developed. Results: Among 124 pts, ctDNA detected ALK variant in 53% (n = 66), mostly EML4::ALK (V1 n = 25, V3 n = 24, V2 n = 7, V5 n = 4). co-GAs were present in 46%, including suppressor gene alt. in 35%. High dNLR was observed in 34% and hypoalbuminemia in 21%. In multivariable analysis, high dNLR (HR 1.67, 95% CI 1.02–2.73; p = 0.04), low albumin (HR 0.34, CI 0.20-0.57; p < 0.001) and suppressor co-GAs (HR 2.24, CI 1.32-3.79; p = 0.003) were independently associated with shorter PFS. Non-V1 variants showed a trend toward inferior PFS (HR 1.73; p = 0.05). For IC-PFS, high dNLR (HR 1.52; p = 0.03), brain metastases (HR 2.98; p < 0.001), and suppressor co-GAs (HR 2.41; p = 0.002) remained independently prognostic. The ALFA score stratified pts into low- (59%), intermediate- (27%), and high-risk (14%) groups, with median PFS of 35.5, 11.1, and 5.5 months (mo.), respectively (p < 0.0001; c-index 0.72). A similar separation was observed for intracranial PFS (IC-PFS; p < 0.0001), with median IC-PFS not reached in the low-risk group, 21.1 mo in the intermediate-risk group, and 5.5 mo. in the high-risk group (c-index 0.72). Overall survival (OS) showed consistent and significant discrimination across ALFA risk groups (log-rank p < 0.0001) with a median OS NR in low-risk, 44.3 mo. in intermediate-risk, and 19.3 mo. in high-risk pts. Conclusions: The ALFA score integrates baseline ctDNA-derived molecular features with routine blood-based parameters to robustly stratify clinical outcomes in ALK + pts treated with frontline brigatinib. This composite score identifies a subset of patients with durable systemic and intracranial benefit from second-generation ALK inhibitors, supporting its potential role as a practical risk stratification tool in routine clinical practice.