Macrophage and iron-associated transcriptomic patterns across human atherosclerotic plaque phenotypes

Macrophage and iron-associated expression patterns are prominent in atherosclerotic plaques, but their reproducibility across lesion phenotypes and dependence on cellular composition remain uncertain. Six human plaque transcriptomic datasets comprising 156 analyzed tissue samples and 104 within-cohort biological units were analyzed using cohort-specific models, paired comparisons, random-effects synthesis, and macrophage-abundance proxy adjustment. Single-cell resources supported cellular localization and patient-level sensitivity analyses. Candidate-ranking sensitivity and leave-cohort-out phenotype discrimination were evaluated. Sixteen transcripts were examined in stimulated macrophage, endothelial, and smooth-muscle-cell cultures. Across five whole-tissue cohorts, pooled standardized effects were 1.63 (95% confidence interval [CI], 0.69–2.57; I²=63%) for the macrophage module and 1.10 (95% CI, 0.78–1.43; I²=0%) for the iron module. In the intraplaque-hemorrhage cohort, adjustment for a plaque-reference macrophage score attenuated the corresponding phenotype coefficients to 0.13 (95% CI, − 0.15–0.41) and 0.77 (95% CI, − 0.18–1.72). Single-cell profiles localized candidates to immune and vascular compartments, whereas external patient-level differences did not survive multiple-testing correction. Held-out areas under the curve ranged from 0.694 to 1.000; perfect discrimination in four paired patients had an exact permutation P = 0.125. Cell cultures showed lipid- and heme-responsive transcripts, with endothelial and smooth-muscle responses that were not uniformly concordant with bulk-tissue directions. Macrophage and iron-associated tissue signals recur across plaque phenotypes and attenuate after adjustment for macrophage-abundance proxies. These patterns nominate context-dependent candidates for further tissue-based evaluation.