Quantifying evolutionary dynamics, heterogeneity, and drug resistance in oncogene-addicted advanced non-small cell lung cancers

Abstract

Non-small cell lung cancer (NSCLC) remains one of the leading causes of cancer-related mortality worldwide. In oncogene-addicted subsets driven by genomic alterations such as EGFR activating mutations, ALK rearrangements, and gene fusions involving RET, ROS1, or NTRK, tyrosine kinase inhibitors (TKIs) lead to rapid initial responses. However, most patients ultimately relapse due to metastatic progression and acquired drug resistance. These disease recurrence and treatment failures underscore a fundamental biological challenge: tumors evolve under therapeutic pressure. This dissertation investigates the genomic, transcriptomic, and evolutionary dynamics that shape drug resistance and tumor heterogeneity in oncogene-driven advanced NSCLC. This thesis specifically leverages one of the largest datasets that combines longitudinal biopsies and autopsies across molecular subtypes of lung cancer, which combined both longitudinal biopsies and a multi-site rapid autopsy cohort. The first chapter of this thesis reviews the biological foundations and clinical contexts necessary to understand these scientific challenges. It synthesizes cancer biology, principles of cancer evolution, tumor heterogeneity, and metastatic dissemination, with emphasis on oncogene addiction in NSCLC. This chapter also outlines current treatment paradigms, including targeted therapy with TKIs and immunotherapy, and summarizes known mechanisms of intrinsic and acquired resistance to these drugs, thereby defining the key knowledge gaps motivating this work. The second chapter characterizes the genomic and transcriptomic landscape of advanced NSCLC. Using a large datasets comprising 1,253 sequencing profiles, including whole-exome (WES) and whole-genome (WGS) sequencing, and bulk RNA-seq, generated from 986 biospecimens across 104 patients, it defines the patterns of somatic mutations, copy-number alterations, gene fusions, mutational signatures, gene expression, immune landscape in this cohort. These analyses establish the genomic and transcriptomic, context within which advanced NSCLC evolved and how therapeutic resistance emerge. The third chapter examines transcriptional programs and tumor microenvironment states using bulk RNA sequencing–based analyses. It integrates immune deconvolution, differential expression analysis, gene set enrichment, and pathway-level interpretation to assess putative off-target mechanisms of resistance such as epithelial-to-mesenchymal transition (EMT) as well as identifying novel therapeutics vulnerabilities like receptor tyrosine kinase signaling activity, and antibody–drug conjugate (ADC) target expression. This chapter also characterizes immune heterogeneity across lesions, distinguishing “immune-hot” and “immune-cold” regions from infered immune cell populations, and evaluates how targeted therapy and immunotherapy affects the immune landscapes of NSCLC. Together, these analyses define functional heterogeneity in the tumor microenvironment that is not fully captured by genomic alterations alone, described in chapter 2. The fourth chapter reconstructs clonal evolution and identifies genomic mechanisms of acquired resistance in EGFR- and ALK-driven NSCLC. By analyzing longitudinal biopsies and multi-site rapid autopsies, and through phylogenetic reconstruction, it distinguishes truncal (clonal) from branch-specific (subclonal) events, identifies convergent and parallel resistance pathways, and maps the timing of resistance emergence relative to therapy exposure using serial biopsies. This chapter demonstrates that, in NSCLC, drug resistance to TKIs can arise through multiple coexisting adaptive mechanisms within the same tumor, rather than a single dominant pathway. The fifth chapter quantitatively measures intra- and inter-tumoral heterogeneity and evaluates their role in shaping metastatic dissemination and therapeutic resistance and adaptation. Using variant allele frequency (VAF) distributions, cancer cell fraction (CCF) modeling, clonal structures from phylogenetic reconstructions, and statistical metrics of tumor diversity, it characterizes clonal diversity within (intra-tumor) and between (inter-tumor) lesions, infers metastatic seeding patterns, and relates tumoral heterogeneity to potential clinical outcomes. The sixth and final chapter synthesizes findings across all five chapters and findings from the genomic, transcriptomic, and evolutionary analyses. It discusses the biological and clinical implications of this work, highlights methodological innovations, and outlines future directions and forward looking statement for evolution-guided precision oncology. Collectively, this dissertation establishes a multi-scale framework for understanding how oncogene-addicted NSCLC adapts under therapy and provides quantitative foundations for designing more durable treatment strategies for patients.