Association of lymph node examination during resection with overall survival (OS) and healthcare costs in patients with early non-small-cell lung cancer (eNSCLC)

e20567 Background: To evaluate the effect of LN examination status on OS and healthcare costs in patients (pts) with eNSCLC. Methods: Retrospective observational study of pts with resected stage IA-IIIB NSCLC (AJCC 7th ed) from SEER data linked with Medicare claims. Eligible pts were ≥65 years at diagnosis (Jan 2010-Dec 2017), had surgery ≤1 month before or 12 months after diagnosis and were continuously enrolled in Medicare Parts A and B for ≥6 months before diagnosis and in Parts A, B and D up to 6 months post-surgery or date of death. Pts were analyzed by LN examination status: no examination (pNX), examination and no LN metastasis (pN0) and examination with N1 (pN1) or N2 (pN2) metastasis. Extended Cox proportional hazards models presented OS hazard ratios (HRs) for specific time periods. Total healthcare costs per patient per month (PPPM) were analyzed using a gamma-log regression model. Results: A total of 9448 pts were included: 996 pNX (11%), 6457 pN0 (68%), 1003 pN1 (11%) and 992 pN2 (10%). pNX pts were slightly older with lower median income, more comorbidities and more baseline smoking history vs other LN examination status cohorts. Stage distribution in pNX and pN0 pts appeared similar. Grade 3/4 tumors were less prevalent in pNX (n = 301, 30%) and pN0 (n = 1934, 30%) pts relative to pN1 (n = 482, 48%) and pN2 pts (n = 424, 43%; P< 0.001). Unadjusted OS was worse for pN1 and pN2 vs pN0 pts and was similar between pNX and pN2 pts. Adjusted OS showed higher risk of death for pN1 and pN2 vs pN0 pts; this risk decreased over time. pNX pts also had higher risk of death vs pN0 pts. Adjusted marginal mean follow-up costs were lowest for pN0 and highest for pN2 pts (All P < 0.001). Costs were higher, but not statistically significant ( P= 0.591) in pNX vs pN0 pts. The majority of total healthcare costs (62-65%) were due to inpatient services. Conclusions: Stage distribution in pNX pts was more similar to that of pN0 than pN1/2 pts, likely due to understaging from lack of LN examination. However, prognosis in pNX pts was nearly identical to that of pN2 pts, suggesting missed LN metastasis diagnosis in a substantial proportion of pNX pts. Despite worse OS in pNX vs pN0 pts, healthcare costs in pNX pts were not significantly higher, which may reflect underutilization of systemic therapy in pts with resected eNSCLC. [Table: see text]