Independent burden and functional status trajectories define prognostic phenotypes in stage IV NSCLC

Abstract Background Prognostic assessment in advanced non-small cell lung cancer (NSCLC) relies on static measures of metastatic burden and ECOG performance status. Functional decline is assumed to be correlated with disease progression, implying these trajectories may be partly redundant. Whether they provide independent prognostic information has not been tested at scale. Methods Using the MSK-CHORD real-world dataset, which integrates longitudinal tumor burden and ECOG status, we identified 1,486 stage IV NSCLC patients with both trajectories at a pre-specified 6-month landmark. Trajectory worsening was tested for statistical independence (chi-square, equivalence testing). Additive prognostic value was assessed using cross-validated Cox models. Four concordance-based risk groups were compared by Kaplan-Meier and adjusted Cox analyses. Results Burden and ECOG trajectory worsening were statistically independent (phi= -0.033; TOST P = 0.005; BF₀₁=17.3), with 47.2% discordant. Each trajectory contributed independent prognostic information (ΔC ≈ 0.02; combined ΔC = + 0.054). Concordance defined four groups with median post-landmark survival from 16.2 months (both worsened; adjusted HR 2.16, P < 0.001) to 32.0 months (both stable). Independence held across landmarks and clinical subgroups. Conclusions Burden and functional trajectories capture distinct disease dimensions; their discordance might distinguishe reversible functional decline from concordant progression, informing treatment versus supportive-care decisions.