Answers are grounded in this paper and cite their sources.
To compare D-dimer, albumin, neutrophil-to-lymphocyte ratio (NLR), haemoglobin and the composite DANH index against stage, metastasis and thrombosis in lung cancer. Retrospective cross-sectional analysis (01/2024–06/2026) of 1,126 cancer patients, 197 with lung cancer; 163 complete cases formed the primary set (14, 44 and 12 minority-class events). DANH-O (0–21) grades each marker into four ordinal strata at pre-specified boundaries; DANH-B (0–4) dichotomises at in-sample Youden cut-points. Sixteen markers and indices were compared by AUC and bidirectional nested modelling (likelihood-ratio test, Benjamini–Hochberg-corrected). Only D-dimer tracked all three outcomes (AUC 0.874, 0.770, 0.772); NLR and albumin tracked stage alone, haemoglobin stage and metastasis. DANH-O never exceeded D-dimer (0.870, p = 0.90; 0.721, p = 0.14; 0.619, p = 0.011, q = 0.046); the reverse test was positive in 1 of 27 comparisons and not robust. DANH-O had higher AUCs than DANH-B across all three outcomes; Youden cut-points varied fourfold across 1,000 resamples (0.050 vs. + 0.001 optimism). Informational value concentrates in the prothrombotic axis. Combination benefit belongs to the index–outcome pair, not the index: it emerges for multi-axial questions and becomes dilution when the outcome is single-axis. For thrombosis, D-dimer is itself a fibrin degradation product, a near-tautological association.
No takes yet. Share an insight, caveat, or question.
Van et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: