Quantifying Early-Stage Lung Adenocarcinoma Progression with a Radiomic Trajectory

Abstract

Determining tumor progression status is critical for early-stage lung adenocarcinoma (esLUAD) diagnosis and treatment, yet histopathology-based grading often overlooks heterogeneity within grades. We propose RadioTrace, a deep contrastive learning framework integrating radiomic and pathological information to learn a radiomic trajectory for quantifying esLUAD progression. Across four multi-institutional cohorts, RadioTrace well predicted tumor phenotypes including spread through air spaces (STAS) and lymph node metastasis (LNM). Survival analyses demonstrated it as an independent prognostic factor (log-rank test p < 0.004 across all cohorts). Within the same pathological grade, it revealed significant survival heterogeneity (p < 0.02 across all cohorts), underscoring the limitations of current grading criteria. Genomic and transcriptomic analyses confirmed associations with progression-related molecular features. Longitudinal analysis of patients with multiple CT follow-ups further showed consistency with continuous progression. These findings demonstrate that RadioTrace enables quantitative, interpretable assessment of esLUAD progression, providing insights beyond histopathology and assisting clinical decision-making.

Fig. 1. Study design and method diagram.

a Illustration of patient cohorts; b workflow of model training with contrastive learning; c workflow of trajectory construction and pseudo-progression score (PPS) inference.

Fig. 2. The radiomic trajectory of esLUAD progression and the distribution of corresponding pseudo-progression score (PPS).

Left column: Tumor image embeddings of multi-institute patient cohorts in the PC space along the RadioTrace. Right column: Distribution of the inferred pseudo-progression score of each tumor. The PPS is in accordance with the pathological grade records. a, b The GDPH1 cohort, in which the radiomic trajectory was constructed; c, d the GDPH2 cohort; e, f the GDPH3 cohort; g, h the SZPH cohort.

Fig. 3. Association between PPS with tumor properties and patient prognosis in all esLUAD tumors and in grade IAC-II.

PPS is significantly higher for tumors with spread through air space (STAS) and lymph-node metastasis (LNM). The disease-free survival (DFS) is significantly better for patients with lower PPS. a–e All esLUAD patients. f–i patients within grade IAC-II.

Fig. 4. RadioTrace is associated with gene expression of esLUAD tumors.

a Distribution of PPS between tumors with and without mutation of certain genes. b Enriched Hallmark gene sets identified by GSEA for tumors groups divided by the RadioTrace. c The correlation between gene expression and PPS of all tumors in the GDPH3 cohort. d Correlation between WGCNA gene modules with PPS. e The identified gene module is associated with patient’s prognosis. * indicates a p-value less than 0.05, ** indicates a p-value less than 0.01, and *** indicates a p-value less than 0.001.

Fig. 5. Longitudinal changes of individual tumors from three patients along with the RadioTrace.

a Case 1 (timely intervention); b Case 2 (potential overtreatment); c Case 3 (missed surgical cured opportunity).