Lung cancer is a leading cause of cancer-related deaths globally. Understanding the biological mechanisms driving its aggressive behavior is crucial. Intratumor heterogeneity (ITH) offers insights into tumor phylogeny, but previous studies were limited by small sample sizes. The TRACERx study, with its prospective design, addresses this limitation by analyzing a larger cohort of patients. The study aims to decipher lung cancer evolution and determine the relationship between ITH and clinical outcomes, particularly focusing on non-small cell lung cancer (NSCLC). Previous research using the first 100 patients from TRACERx showed pervasive genomic ITH and a significant association between somatic copy number alteration (SCNA) heterogeneity and poor prognosis. This study expands upon this research by including data from the first 421 patients, increasing the statistical power for analysis. By leveraging multiregion exome primary tumor data, the research investigates the relationship between established and new measures of ITH and clinical outcomes to further our understanding of NSCLC evolution. The 421 patients in this cohort were recruited from 19 hospital sites in the United Kingdom and represent a range of NSCLC subtypes and stages.

Previous studies using multiregion sequencing provided valuable insights into intratumor heterogeneity (ITH) and branched evolution in cancers. However, these studies were limited in sample size, typically including less than 100 patients for a given cancer type. This limited statistical power for drawing robust conclusions about genomic and clinical analyses. The functional relevance of ITH in determining clinical outcomes, particularly in relation to personalized medicine, remained a subject of debate. The TRACERx study was initiated to address these limitations by using a prospective, multicenter design with a significantly larger cohort of patients. A previous analysis of the first 100 TRACERx patients demonstrated widespread genomic ITH and an association between SCNA heterogeneity and prognosis, but no relationship between mutational ITH and outcome was observed. This current study builds on these findings with a much larger dataset.

The TRACERx study prospectively recruited 421 patients with early-stage NSCLC across 19 UK hospital sites. The cohort included 248 lung adenocarcinomas (LUAD), 138 lung squamous cell carcinomas (LUSC), and 46 other NSCLC subtypes. 1,644 tumor regions were subjected to whole-exome sequencing (WES). The study employed various bioinformatics methods, including CONIPHER for phylogenetic tree reconstruction, ParallelGDDetect for WGD detection, and dNdScv for quantifying selection. Mutational signatures were extracted using a hierarchical Dirichlet process model. The study also incorporated clinical data, including smoking history and treatment information. To validate their phylogenetic reconstruction approach, the researchers developed a simulation framework that reproduced specific features of the tumors and sequencing data in the TRACERx 421 cohort. This framework was then used to benchmark their approach against existing methods. The study used various statistical methods, including Pearson's r, Fisher's exact test, and Cox proportional hazards models, to analyze the data. The clonality of somatic mutations was classified based on phylogenetic cancer cell fraction (PhyloCCF). Copy number alterations were detected using ASCAT and Sequenza, and the weighted genome instability index was calculated.

The study revealed that in LUAD, mutations in 22 out of 40 common cancer genes were under significant subclonal selection. 8% of LUADs from ever-smokers lacked evidence of tobacco-induced mutagenesis, suggesting alternative tumorigenic mechanisms. Large subclonal expansions were associated with positive subclonal selection and shorter disease-free survival (DFS). Subclonal, but not truncal, WGD was associated with shorter DFS. Copy number heterogeneity was associated with extrathoracic relapse within 1 year post-surgery. Significant context dependency was observed between truncal and subclonal events. The likelihood of subclonal alterations was influenced by the presence of truncal alterations in the same gene (e.g., TP53). There was a significant association between high SCNA ITH and shorter DFS, specifically associated with early (<12 months) and extrathoracic relapse. Other evolutionary metrics, such as subclonal WGD and recent subclonal expansion, better predicted the overall likelihood of relapse. Recent subclonal expansion score remained a significant predictor of DFS in a multivariable model, even after adjusting for standard clinical indicators.

This large-scale study provides strong evidence for the importance of subclonal selection and expansion in driving lung cancer progression and poor prognosis. The finding that 8% of LUADs from ever-smokers lacked smoking-related mutations suggests alternative mechanisms of tumor initiation, warranting further investigation. The identification of evolutionary dependencies between different genomic events highlights the complexity of lung cancer evolution. The study's strength lies in its large sample size and prospective design, which provide greater statistical power than previous studies. The use of multiregion sequencing provided a more comprehensive understanding of ITH and its clinical implications than would be possible with single-region sampling. The findings have important implications for personalized medicine by suggesting potential prognostic biomarkers and treatment targets. Future research could focus on exploring these alternative mechanisms and the interplay of different genomic events to improve lung cancer treatment and outcomes.

This study, using the TRACERx cohort of 421 patients with early-stage NSCLC, provides a comprehensive analysis of lung cancer evolution. It demonstrates the frequent occurrence of subclonal selection and the association of recent subclonal expansions with poor prognosis. The identification of LUADs in smokers lacking smoking-related mutations highlights the need to investigate alternative tumorigenic mechanisms. The results underscore the importance of ITH in predicting clinical outcomes and suggest that multiple evolutionary metrics should be considered for improved prognostication. Future studies should focus on validating these findings in larger independent cohorts and investigating the mechanistic basis of these observations.

The study is limited to a UK cohort, and the generalizability of the findings to other populations may need further investigation. The reliance on self-reported smoking history could introduce some degree of bias. While the study included a large number of patients, it still only represents a subset of all lung cancer patients, and there may be other factors impacting prognosis not captured in the analysis. Although the study accounts for several factors when building its model, there may be additional unmeasured confounders that could influence the results.