Age-vulnerability interaction and real-world treatment outcomes in EGFR -mutant non–small cell lung cancer.

e20685 Background: Older adults with EGFR-mutant metastatic non-small cell lung cancer (NSCLC) are underrepresented in clinical trials, and treatment decisions often rely on chronological age rather than biologic reserve. Whether vulnerability modifies the relationship between age and real-world treatment outcomes remains unknown. Methods: We conducted a retrospective cohort study of patients with EGFR-mutant metastatic NSCLC treated with first-line EGFR TKI from 2012-2025 at Northwell Health. Baseline vulnerability was defined using a composite score incorporating 5 established risk factors: high comorbidity burden, performance status (ECOG≥2), hypoalbuminemia ( < 3.5 g/dL), polypharmacy (≥5 medications), and recent hospitalization within prior 6 months. Patients were classified as non-vulnerable (0-1) vs vulnerable (≥2). Outcomes included time to treatment discontinuation (TTD) and progression-free survival (PFS). Multivariable Cox proportional hazards models adjusted for diagnosis era, EGFR mutation subtype, and baseline brain metastases. Effect modification between age (≥75 vs < 75) and vulnerability was tested using interaction terms. Results: Among 112 patients (median age 72, range 40-90), 37% (n = 41) were aged ≥75. Vulnerability (score≥2) was present in 50% of younger and older patients (p = 0.3). Patients received osimertinib (52%), erlotinib/gefitinib (33%), or afatinib (14%). In adjusted models, age ≥75 alone was not associated with TTD (HR 0.88, 95% CI 0.49-1.56, p = 0.65). However, a significant age-vulnerability interaction was noted (p = 0.03). Among non-vulnerable patients (n = 51), age ≥75 was associated with longer TTD (HR 0.43, 95% CI 0.20-0.95) while among vulnerable patients (n = 39), age ≥75 predicted shorter TTD (HR 2.60, 95% CI 1.07-6.22). A similar interaction was observed for PFS (p = 0.04). Among non-vulnerable patients, age ≥75 was not associated with PFS, whereas among vulnerable patients, age ≥75 predicted a shorter PFS (HR 2.1, 95% CI 0.5-10). Diagnosis era, EGFR subtype, and baseline brain metastases remained significant predictors in multivariable models. Conclusions: Chronological age alone does not explain treatment outcomes in EGFR-mutant NSCLC. Vulnerability metric, derived from 5 readily available clinical factors, modifies how age translates into attrition and progression, identifying a high-risk subgroup of older vulnerable patients. Fit older adults achieve outcomes comparable to younger patients. These findings advance precision geriatric oncology by moving beyond one-size-fits-all age cutoff and support vulnerability-informed treatment sequencing and warrant prospective validation.