8525 Background: Resected early-stage NSCLC has high risk of recurrence, and tyrosine kinase inhibitors (TKIs) and immune checkpoint inhibitors (ICI), each requiring testing for precision biomarkers, have recently been approved in the adjuvant (adj) setting. We assessed the potential value of multi-gene testing in early NSCLC via enabling timely 1L therapy selection at recurrence and through avoidance of adj ICI in pts with driver alterations unlikely to benefit per current guidelines and the label for atezolizumab in first line (1L) NSCLC. Methods: Using the nationwide (̃280 US cancer clinics) de-identified EHR-derived rw Flatiron Health- Foundation Medicine clinico-genomic database, we selected 6,725 evaluable pts with LUAD who underwent tissue comprehensive genomic profiling (CGP) as part of routine cancer care (01/2011 – 06/2021). We focused on alterations in 4 drivers ( EGFR, ALK, ROS1, RET) and studied prevalence in early-stage specimens versus advanced (adv) specimens, as well as the rate of timely delivery of 1L therapy at recurrence for pts receiving CGP in early disease. We estimated the cost implications of adj ICI in pts with PD-L1+ LUAD and an ALK, ROS1, or RET fusion with a probabilistic decision tree. Results: CGP was performed on 1,490 specimens collected prior to adv disease (stage I 36%, II 27%, IIIA 37%) and ordered prior to adv diagnosis for 981 pts (15% of total, median 10 weeks after initial diagnosis). In specimens collected in early (n = 1,490) or adv (n = 5,130) stage LUAD, CGP identified drivers in EGFR (early/adv: 13%/16%), ALK (2.0%/4.2%), ROS1 (0.7%/1.1%), RET (1.0%/1.1%), and prevalence was similar when limiting to PD-L1+ cases. Studying 596 pts with recurrence and CGP on samples collected prior to recurrence, pts with CGP results obtained before (n = 196) vs after (n = 400) recurrence had less time between recurrence and start of any 1L therapy (median 3.1 vs 5.9 weeks, p < 0.001). In the subset with a targetable driver detected, 32/42 (76%) with CGP before recurrence initiated matched 1L TKI while 43/67 (64%) with CGP after recurrence received matched 1L TKI (p = 0.2). Through avoidance of ICI in PD-L1+ early LUAD with an ALK/ ROS1/ RET driver, we estimate universal CGP (compared to EGFR single-gene testing) could reduce per pt expected costs by $875 (an average incremental $4,050 treatment cost reduction compared to a $3,175 increase in diagnostic cost per pt). Conclusions: CGP of early-stage LUAD can identify EGFR, ALK, ROS1 and RET drivers and enable appropriate selection of precision therapies and timely use of effective 1L therapy at recurrence. Assuming adj ICI maintains the lack of activity in RET/ ROS1 as seen for ALK, CGP could represent a cost-effective approach for avoiding futile adj ICI and reducing the risk of subsequent TKI-associated toxicity. Additional analysis of driver+ pts receiving adj ICI is needed to help balance the risk/benefit for these high risk pts.