Abstract B027: A global analysis of fusion transcripts and recurrence in NSCLC primary-lymph node pairs

Abstract Introduction: In Non-Small Cell Lung Cancer (NSCLC), low-fidelity transcription mechanisms such as ecDNA (extra-chromosomal DNA) are an increasingly relevant aspect of progression and prognosis, warranting comprehensive investigation into stability and prevalence of chimeric RNA transcripts. Here, we describe global distribution of fusion transcripts in encounter-matched primary tumor and lymph node metastases and their prognostic implications. Methods: RNAseq was performed on resected FFPE (encounter-matched primary and lymph nodes) from 121 stage IIB-IIIA (T1-4N1-2M0) NSCLC patients, in addition to 693 Stage IA1-IB cases (T1a-2aN0M0). STAR and STAR-Fusion were used to align data to the GRCh37 genome. Differential gene expression was based on specimen source (primary or lymph node). Aggregate analyses were also performed, including determination of a fusion burden metric, assessing the relationship between primary tumors, lymph node metastases, and recurrence. Deconvolution of Tumor Microenvironment was performed via Quantiseq. Results: Comparing primary tumors to paired lymph node metastases, 7,930 ensembl IDs were differentially-expressed at FDR < 0.001 (n=101 pairs). A total of 613 fusions (685 unique transcripts) were detected from paired primaries and lymph node metastases (n = 242). Frequency of specific gene fusions in primaries ranged from 0-0.963 (median 0.019), compared to 0-0.870 (median 0.0093) in LN metastases (p=1.87×10-4). We observed fusion of TULP4-RP11-732M18.3 as pervasive in NSCLC, occurring in 84-87% of primary tumors and lymph nodes, with a 91% retention rate. Conversely, fusions involving CCDC7 or SEPT14 were preferentially observed in metastases, where 82% of CCDC7 fusions and 75% of SEPT14 fusions were not observed in the primary tumor despite occurring at high frequency (66% and 18.5%, respectively) among LN metastases. Interestingly, median fusion burden (unique gene-gene fusions) for primaries was 14 compared to 11 (p=2.06×10-7) in LN metastases (median change -4). Among N0 cases that did not receive adjuvant therapy, median fusion burden in non-recurrent tumors was 15 (n=570), compared to 13 in those that recurred over 3-60 months (n=123; p = 0.04). Interestingly, there was no significant difference in fusion burden between recurrent cases and primary tumors with LN metastases (n = 108). For both recurrent cases and LN-paired primaries, the difference in fusion burden compared to LN metastases (n = 110) was significant (p=2.10×10-4 and 2.06×10-7). A subset of LN metastases (n=27) increased in fusion burden relative to the primary tumor; these primaries uniquely exhibited 33% decrease in regulatory T cell content (p=0.008), 20% decrease in M2 macrophage content (p = 0.01 with M1 macrophages static), and 40% decrease in B-cell content (p=0.05). Conclusions: Fusion transcripts are pervasive in RNAseq data, suggesting a broader and more variable landscape than accounted for by primary tumors alone. Concurrent ongoing studies explore the role of ecDNA in fusion burden changes related to NSCLC progression and metastasis. Citation Format: Kelly M. Cagin, Keri L. Denson, Sierra R. Broad, Wara M. Naeem, Arsalan M. Khan, Michael M. Liptay, Christopher W. Seder, Jeffrey A. Borgia. A global analysis of fusion transcripts and recurrence in NSCLC primary-lymph node pairs [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Fusion-Positive Cancer: From Discovery to Therapy; 2026 Jan 13-15; Philadelphia PA. Philadelphia (PA): AACR; Cancer Res 2026;86(1_Suppl):Abstract nr B027.