Multi-omic profiling separated recurrence risk within pathological stage in resected EGFR-mutant lung adenocarcinoma

A median concordance index of 75.4% across four validation cohorts

The model was built from clinical, genomic and transcriptomic data together and tested in one internal and three external cohorts, and the authors describe its performance as superior and reproducible. A concordance index measures how often two patients are ranked in the correct order for recurrence, so 75.4% describes discrimination and not an absolute risk for any patient. The abstract gives the median alone: no value for any single cohort, no confidence interval, and no figure for the stage-based approach the model is offered against.

TP53, copy number alterations and APOBEC signatures rose with stage, and RBM10 clustered with L858R

Genomic instability tracked with stage. TP53 co-mutations, copy number alterations and APOBEC-associated mutational signatures all increased with pathological stage, so they carry information that stage already carries. RBM10 co-mutations were enriched in tumours with L858R mutations and correlated with upregulation of WNT signalling and epithelial-mesenchymal transition, a correlation within this cohort with no recurrence figure attached to it. Transcriptomic features predicted recurrence more accurately than the clinical or genomic variables alone, although no figure is reported for those single-modality models.

Stratification inside stage I, against an adjuvant standard set by pathology alone

Adjuvant osimertinib is the standard of care for resected stage IB to IIIA EGFR-mutant disease following ADAURA, and adjuvant decisions are guided by pathological stage alone. The authors note that real-world data show up to 40% of patients remain disease-free at five years without it. The model stratified risk within individual stages, including stage I, and the authors state that it identified the patients most likely to benefit from adjuvant EGFR TKI. No trial is reported in which treatment was assigned on the basis of the model, so that benefit is a prediction about who is at risk rather than a measured effect of giving or withholding the drug.

“These findings define the molecular heterogeneity of early-stage EGFR-mutant LUAD and support multi-omic risk stratification to inform adjuvant EGFR TKI decisions beyond pathological stage.”Saw et al., Journal of Thoracic Oncology (2026)

The authors state that prospective validation in larger cohorts will be required to confirm the findings.