The Global Diversity Gap: Most large-scale cancer genomics studies have enrolled predominantly White patients. Lung adenocarcinoma (LUAD) affects populations worldwide with differing risk factor profiles - Asian female never-smokers with EGFR mutations, East Asian smokers with KRAS mutations, and Western smokers with distinct carcinogen exposure patterns - requiring truly diverse study designs.
Proteogenomics as an Integrative Platform: This study applied proteogenomics - the combined analysis of genomics, transcriptomics, proteomics, and phosphoproteomics - to 406 LUAD tumors from participants spanning diverse ethnicities in the CPTAC and ICPC consortia. This multi-omics integration reveals molecular features invisible to any single platform.
Study Scale and Design: 406 tumors were profiled across all molecular layers. The cohort included patients from the US, Taiwan, Japan, and other sites, with detailed information on smoking history, tumor stage, driver mutations, and treatment outcomes. This is among the largest and most diverse LUAD proteogenomic studies to date.
Key Questions: The study asked how carcinogen exposure histories shape mutation signatures, how molecular subtypes stratify survival, and whether shared therapeutic vulnerabilities exist across diverse patient populations that could guide treatment.
NMF-Based Classification: Non-negative matrix factorization (NMF) clustering of integrated proteogenomic data identified four molecularly distinct LUAD subtypes. These clusters were defined by co-variation of protein expression, mutational patterns, and chromosomal instability metrics across the 406 tumors.
Unstable Proliferative (UP) Subtype: Characterized by high chromosomal instability, TP53 mutations, and aggressive proliferative signaling. This subtype had the poorest survival outcomes and was enriched in smokers with heavy carcinogen exposure and KRAS co-mutations.
Quiescent EGFR (QE) Subtype: Dominated by EGFR mutations, relatively low chromosomal instability, and a quiescent transcriptional state. This subtype was enriched in Asian female never-smokers and associated with good EGFR TKI response, consistent with clinical experience.
Immune Active KRAS (IAK) and Stable Early Stage (SES) Subtypes: The IAK subtype showed high immune cell infiltration and KRAS co-mutations, suggesting potential immunotherapy responsiveness. The SES subtype represented stable, early-stage tumors with favorable prognosis and lower therapeutic urgency, though a subset of stage I SES-adjacent tumors showed unexpectedly poor prognosis (see late-like early-stage finding).
Two Distinct Carcinogen Classes: The study identified two major carcinogen mutation signature clusters across LUAD tumors. One cluster was driven by nitrosamine compounds (tobacco-specific nitrosamines such as NNK), while the other reflected polycyclic aromatic hydrocarbons (PAHs) and nitro-PAHs from combustion.
Nitrosamine Signatures in Never-Smokers: Surprisingly, the nitrosamine-dominated signature was enriched in Asian female never-smokers, particularly those with EGFR mutations. This suggests that dietary, occupational, or air quality exposures to nitrosamines - not just cigarette smoking - drive LUAD mutagenesis in this high-risk demographic.
Nitro-PAH and PAH Signatures in Smokers: Heavy smokers showed predominant PAH and nitro-PAH signatures, consistent with combustion-derived carcinogens from cigarette smoke. These signatures were associated with KRAS mutations, SMARCA4 alterations, and the Unstable Proliferative subtype.
Carcinogen-Driven Therapeutic Implications: Understanding which carcinogens drove mutagenesis helps explain why EGFR mutations predominate in never-smokers and why KRAS mutations dominate in smokers. It also suggests that preventive strategies in high-risk Asian female populations should consider reducing non-cigarette nitrosamine exposures.
IGF2BP3 as a Genomic Instability Marker: IGF2BP3 (Insulin-like Growth Factor 2 mRNA Binding Protein 3) emerged from the proteogenomic analysis as a key protein elevated in tumors with high chromosomal instability. It was identified as a biomarker of aggressive, genomically unstable LUAD with poor prognosis.
BIC: Breakage Intensity Clustering Metric: The study introduced a novel metric called Breakage Intensity Clustering (BIC) to quantify the extent and pattern of chromosomal structural rearrangements in each tumor. BIC integrates copy number breakpoints and structural variant density to provide a single chromosomal instability score.
BIC and Survival: High BIC scores correlated strongly with worse recurrence-free survival and overall survival, independent of stage and driver mutation status. Tumors in the Unstable Proliferative subtype had the highest BIC scores, linking structural genomic instability to the most lethal LUAD biology.
Late-Like Early-Stage Subtype: A critical finding was the identification of a subset of proteomically-classified stage I patients (36.4% of the C2 proteomic cluster) with gene expression and protein profiles resembling late-stage disease. These patients had the worst recurrence-free survival among all stage I patients, demonstrating that proteomics detects aggressive biology that pathological staging misses.
Five Tiers of Therapeutic Targets: The study constructed a systematic therapeutic vulnerability atlas organized into five tiers based on evidence strength: tier 1 (approved targeted therapies), tier 2 (biomarker-matched trials), tier 3 (functional phosphoproteomic targets), tier 4 (kinase enrichment targets), and tier 5 (metabolic and immunologic vulnerabilities).
Kinase Enrichment Analysis: Phosphoproteomic data enabled kinase activity enrichment analysis, identifying which kinases were hyperactivated in each molecular subtype. This approach revealed druggable kinases beyond classical EGFR and ALK, including Aurora kinase and CDK family members, particularly in the Unstable Proliferative subtype.
Subtype-Matched Drug Candidates: Each of the four NMF subtypes had distinct enriched kinase activities, enabling subtype-specific drug matching. For example, the Unstable Proliferative subtype showed Aurora B kinase hyperactivation, suggesting sensitivity to Aurora kinase inhibitors. The Immune Active KRAS subtype showed enrichment for immune activation pathways addressable with checkpoint blockade.
Cross-Ethnic Drug Targets: Importantly, some therapeutic vulnerabilities were shared across different ethnic and demographic subgroups, suggesting that molecular subtype - rather than ethnicity - should guide drug selection. This challenges race-based prescribing in favor of biology-driven treatment matching.
Cross-Cohort Batch Effects: Combining proteogenomic data from multiple centers and countries introduces potential batch effects in protein quantification and sequencing. While computational harmonization was applied, residual technical variability may affect subtype assignments at boundaries.
Snapshot Multi-Omics: All molecular profiling was performed on treatment-naive primary tumors. Proteogenomic landscapes evolve under treatment pressure, and longitudinal sampling - including metastatic biopsies and liquid biopsies - would capture resistance mechanisms not seen in this dataset.
Clinical Validation of Subtypes: The four NMF subtypes are molecularly defined but not yet validated as clinical classification entities in prospective treatment trials. Clinical trials specifically enrolling and treating patients by LUAD molecular subtype are needed to confirm that subtype-guided treatment improves outcomes.
BIC and IGF2BP3 Assay Development: Before BIC scoring or IGF2BP3 protein quantification can be used clinically, they need to be converted to practical laboratory assays with validated cutoffs, reproducibility data, and regulatory approval for clinical use in pathology settings.