From imaging heterogeneity to clinical decision: a novel imaging biomarker based on bone marrow kinetics for advanced
Yubo Wang1,2, Zhiheng Yao3, Jingjing Wan4
1Department of Nuclear Medicine, National Cancer Center, National Clinical Research Center for Cancer/Cancer Hospital & Shenzhen Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Shenzhen, China.
Background:
Risk of disease progression varies substantially among advanced non-small cell lung cancer (NSCLC) patients who achieve a partial response to initial therapy, and this subgroup currently lacks practical prognostic tools. This study aimed to develop a novel imaging biomarker by assessing bone marrow kinetic heterogeneity via pretreatment dynamic 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT). We retrospectively analyzed 51 patients with advanced NSCLC, focusing on the 32 who attained a partial response following first-line treatment. Time-activity curves (TACs) from the tumor and bone marrow were metabolically decomposed. Through integrated analysis of clinical characteristics, conventional PET parameters, and 28 TAC-derived features, coupled with a rigorous bootstrap-LASSO-Cox procedure (1000 resamples) for robust feature selection, we developed a multivariate Cox model and a simplified risk-stratification model.
Results:
The bootstrap-LASSO-Cox analysis identified the slope of the bone marrow free FDG TAC between 10 and 30 min post-injection (BM_Cf_Slope10-30) as the most significant independent predictor. In the multivariate Cox model, this feature was strongly associated with a reduced risk of progression (Hazard Ratio = 0.10, 95% CI: 0.03-0.35, P < 0.001) and demonstrated good discriminative ability (C-index = 0.77, 95% CI: 0.68-0.86). Using an optimal cutoff value of -0.01119, this single parameter significantly stratified patients into high- and low-risk groups with distinct progression-free survival (Log-rank P < 0.001). The simplified model showed robust predictive performance, with average time-dependent AUCs of 0.792, 0.774, and 0.731 at 12, 18, and 24 months, respectively, and provided clinical net benefit. Neither conventional metabolic parameters nor clinical characteristics showed significant predictive value.
Conclusion:
For advanced NSCLC patients achieving an initial partial response, pretreatment assessment of bone marrow kinetics via dynamic 18F-FDG PET/CT emerges as a potential prognostic biomarker. The TAC slope feature, which reflects early interstitial clearance, can identify patients at high risk of early progression, offering potential for personalizing surveillance strategies and guiding timely therapeutic interventions.
