Pseudotime-Based derivation of a PET-Based metabolic progression index for prognostic stratification in Extensive-Stage SCLC

To develop and validate a pseudotime-derived imaging biomarker integrating volumetric, metabolic, and spatial information from baseline 18 F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography (18 F-FDG PET/CT) for prognostic stratification in patients with extensive-stage small cell lung cancer (SCLC). This retrospective study included 83 patients with stage IV SCLC who underwent baseline 18 F-FDG PET/CT prior to systemic therapy. Whole-body tumor lesions were semi-automatically segmented, and conventional PET-derived metrics reflecting tumor burden, metabolic activity, inter-lesional heterogeneity, and spatial dissemination were extracted. A diffusion map–based manifold learning framework was applied to derive a latent progression axis, termed the Metabolic Progression Index (MPI), representing a pseudotime-based continuum of metastatic disease severity. The primary endpoint was 12-month all-cause mortality. Prognostic performance of MPI was evaluated using logistic regression, receiver operating characteristic (ROC) analysis, and internal bootstrap validation, and compared with conventional PET-derived parameters. During 12-month follow-up, 60 patients (72.3%) died. MPI showed strong correlations with spatial dissemination and moderate correlations with tumor burden metrics, without being fully explained by any single conventional parameter. In multivariable analysis, MPI emerged as an independent predictor of 12-month mortality (OR per 0.1-unit increase = 1.87, 95% CI: 1.41–2.62; p < 0.001). The MPI-based model demonstrated superior discriminative performance compared with conventional PET-derived models, with significant incremental prognostic value confirmed by likelihood ratio testing. A pseudotime-based integration of baseline 18 F-FDG PET/CT features captures a latent continuum of metastatic disease architecture in extensive-stage SCLC. MPI provides prognostic information beyond conventional PET-derived metrics and may serve as a robust imaging biomarker for risk stratification.