Spatial transcriptomics uncovers patterns of transcriptional reprogramming and immune evasion in small cell lung cancer

Histologic transformation from non-small cell lung cancer (NSCLC) to small cell lung cancer (SCLC) is a major mechanism of therapeutic resistance, yet the spatial and molecular programs governing this phenotypic switch remain poorly defined. 10X Xenium In Situ spatial transcriptomics was used to generate subcellular-resolution transcriptomic maps of 22 primary and metastatic lung tumors across 16 patients with SCLC transformation, totaling >10 million cells. Matched single-cell RNA sequencing enabled integrated analysis of transcriptional states and served as a complementary modality to support spatial findings. Spatial profiling with a 389-gene panel encompassing neuroendocrine (NE), epithelial, stemness, immune, and stromal programs revealed 30 distinct cell types and marked inter- and intra-tumoral heterogeneity across lung adenocarcinoma (LUAD), squamous cell carcinoma (LUSC), and small cell lung cancer subtypes (ASCL1+: SCLC-A and NEUROD1+: SCLC-N). Tumors organized into several canonical structural patterns: (1) discrete subtype compartmentalization, (2) speckled NE transformation, and (3) self-contained, multilayered subtype organization.A representative resection of a lymph-node metastasis with SCLC transformation displayed two separate tumor regions, each with a striking radial arrangement. In the first region, admixed LUAD and NOTCH1- SCLC-A formed the tumor core, followed by concentric layers of NOTCH1+ SCLC-A and SCLC-N, with immune cells at the periphery. Given the established role of Notch signaling in SCLC subtype switching, and our observation of inhibitory DLL3-NOTCH1 interactions across neighboring tumor subtypes, these regulatory relationships may contribute to the spatial compartmentalization of tumor subtypes. The second tumor region lacks the LUAD and NOTCH1+ SCLC-A populations, instead composed of a NOTCH1- SCLC-A core encased by SCLC-N, again surrounded by immune zones. Pseudotime analysis of the Xenium data across both regions also recapitulated a trajectory from LUAD to SCLC-A to SCLC-N. These two findings, combined with previous studies characterizing SCLC-N as an “immune-cold” tumor devoid of immune infiltrate, suggest that this spatial patterning and interplay between subtypes may be an important mechanism for immune evasion and progression in a hostile environment.Due to imbalance in cell number and sample area within the cohort, spatial comparisons were constrained to smaller regions of interest (ROIs). Clustering ROI cell-cell adjacency matrices uncovered recurring spatial interaction patterns associated with specific tumor subtypes and microenvironmental states, such as necrosis or immune infiltration. These analyses highlight distinct architectural modes of neuroendocrine transformation and underscore the role of spatial context in shaping phenotypic evolution during resistance.