Organizational Reconfiguration Across Normal Lung Architecture and Early Lung Malignancy: A Cross-Dataset Spatial Analysis

BackgroundNormal lung function depends on coordinated organization among epithelial, stromal, extracellular-matrix, immune, vascular, mechanical, and environmental components. Malignant transformation may therefore involve not only changes in cellular abundance or molecular activity, but also changes in the spatial organization and relationships among these components. MethodsA previously constructed normal human lung architectural reference containing 44 records was compared with organizational measurements available in two independent lung malignancy datasets. The normal reference contained 15 entity records, 8 environmental records, and 21 relationship records. Before malignant-transition results were attached, each normal record was adjudicated as having a DIRECT, CONCEPTUAL, NO MATCH, or UNRESOLVED correspondence to the cancer measurement framework. Eighteen of 44 normal architectural records had direct or conceptual representation in the malignant datasets. These records mapped to seven unique representative organizational measurements. Malignant-state changes were evaluated independently in (1) a precursor-to-lung-adenocarcinoma comparison in GSE307534 and (2) a cross-sectional EARLY-to-INVASIVE comparison in the Takano/Suzuki cohort. The two datasets were not treated as a single longitudinal sequence. ResultsOf the 44 normal architectural records, 5 had direct measurement correspondence, 13 had conceptual correspondence, 25 had no adequate measurement match, and 1 remained unresolved. The 18 represented normal records mapped to seven unique representative cancer-side organizational measurements. All seven measurements changed in the same direction across the two malignancy comparisons. Five measurements—epithelial, stromal, macrophage/myeloid, T-cell, and extracellular-matrix spatial autocorrelation—showed increased spatial autocorrelation in both comparisons and achieved q ≤ 0.05 in both datasets. Two cross-domain relationships also changed concordantly but did not meet the same cross-dataset statistical-support criterion. Epithelial–ECM association decreased in both datasets, with corrected statistical support in GSE307534 but not in the Takano/Suzuki cohort. Epithelial–macrophage/myeloid association also decreased in both comparisons but lacked corrected statistical support in both datasets. An exploratory patient-level visualization of 22 EARLY and 8 INVASIVE Takano/Suzuki cases showed a descriptive shift toward greater epithelial and broader multicellular spatial autocorrelation in the INVASIVE group. Median standardized epithelial autocorrelation was -0.170 in EARLY cases and +0.874 in INVASIVE cases. Median multicellular-context autocorrelation was -0.458 and +1.066, respectively. ConclusionsAcross two independent lung malignancy comparisons, the most reproducible organizational change was increased spatial autocorrelation across multiple epithelial, stromal, immune, and extracellular-matrix domains. Cross-domain epithelial relationships showed concordant weakening but less consistent statistical support. These findings support further investigation of malignant progression as a change in multicellular spatial organization rather than solely as a change in individual cellular or molecular components. The present analysis does not establish causality, a diagnostic classifier, or a continuous normal-to-malignant trajectory.