Abstract Background: Tumor heterogeneity plays a crucial role in disease progression, metastasis and treatment response in lung adenocarcinoma (LUAD). Limited availability of multiple molecularly profiled samples from the same patients poses a challenge to validate models of tumor evolution using real-world data. We investigated genomic changes in a large cohort of patients with clinically directed DNA sequencing of multiple LUAD samples. Methods: We analyzed longitudinal genomic data from 524 patients with at least two LUAD samples sequenced with MSK-IMPACT- a targeted panel that identifies somatic mutations, copy number alterations and gene fusions in 341-505 genes. Our analysis of 1084 tumors (634 primaries [P] and 450 metastases [M]) included 189 matched P-P pairs, 99 M-M pairs and 271 P-M pairs. Pairs were required to share an identical somatic mutation to confirm relatedness. Additional criteria included: excision of primary tumor prior to excision of the metastasis in P-M pairs collected over 90 days apart and collection of M-M pairs from the same anatomic site. Intra-pulmonary metastases were removed from P-P pairs. Results: Tumor mutational burden was correlated in paired specimens from the same patient, irrespectively of P-P (R= 0.75, p<0.001; Spearman correlation), P-M (R=0.72, p<0.001) or M-M (R=0.6, p<0.001) status. Chromosomal instability, measured as the fraction of genome altered by copy number changes (FGA), was also correlated in P-P (R=0.69, p<0.001) and M-M (R=0.57, p<0.001) pairs, but tended to increase in metastases with respect to matched primaries (median 0.58 vs. 0.72, p<0.001; Wilcoxon test). In M-M pairs, changes in FGA differed by anatomic site. Liver pairs had the strongest correlation (n=17 pairs, R=0.86, p <0.01), whereas brain pairs had the weakest (n=35 pairs, R=0.085, p=0.71). Most somatic mutations were shared between paired samples (P-P: 639/1055, 60%, P-M: 1563/2877, 54%, M-M: 550/958, 57%), including those in LUAD drivers- TP53, KRAS, and EGFR. Conversely, most oncogenic copy number alterations - including CDKN2A deletions and MET amplifications - were private to one specimen (P-P: 183/276, 66%, P-M: 880/1148, 76%, M-M: 430/601, 71%), most often in the later sample (P-P: 113/183, 61%, P-M: 636/880, 72%, M-M: 212/430, 49%). Clinically targetable alterations (OncoKB Level 1) tended to be shared by tumor pairs from the same patient (93% shared vs. 6% private in P-P, p<0.01; 85% vs. 14% in P-M, p<0.01; 78% vs. 21% in M-M, p <0.01). Conclusions: In LUAD, metastatic specimens exhibit increased chromosomal instability in relation to their matched primaries. This translates into unique copy number alterations detected only in the metastasis. By contrast, driver mutations - which account for most of the clinically targetable alterations with currently approved FDA drugs - are more often shared between paired samples from the same patient. Citation Format: Brooke Mastrogiacomo, Elizabeth G. Dunne, Cameron N. Fick, John Nguyen, Jairam Hathwar, Manendra Lankadasari, Yuan Liu, Soo-Ryum Yang, Jason Chang, Walid K. Chatila, Nikolaus Schultz, David R. Jones, Francisco Sanchez-Vega. Longitudinal analyses of clinical sequencing data provide novel insights into the evolutionary dynamics of lung adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2556.