Camryn Bass

Small cell lung carcinoma (SCLC) is the most aggressive common form of lung cancer; patients with limited-stage SCLC have a median overall survival (mOS) of 19.8 months and extensive stage patients have a mOS of 6.5 months. Recent literature suggests transcriptional and immune intertumoral heterogeneity drive outcomes and underlie specific therapeutic vulnerabilities. One system classifies SCLC by molecular biomarkers (ASCL1, NEUROD1, POU2F3, or YAP1/inflamed). While this classification distinguishes group phenotypes, there is expression overlap, immune microenvironment heterogeneity, and the system has failed to show clinical prognostic utility. We hypothesize that clustering integrated survival and transcriptomic data will reveal molecular subgroups that more accurately stratify prognosis in early-stage tumors and facilitate clinical use.
We constructed a retrospective cohort of 49 patients who underwent surgical resection for primary untreated SCLC between 2002 and 2025 at the Institut Universitaire de Cardiologie et de Pneumologie de Québec. Clinical, pathological, and survival outcome data were retrieved from the IUCPQ biobank. Immunohistochemistry for INSM1, ASCL1, NEUROD1, and POU2F3 was tested on tumor tissue microarrays and reviewed by a thoracic pathologist to classify tumors and describe neuroendocrine status. Bulk RNA seq was performed on frozen tumor samples matched with non-tumoral adjacent lung tissue from the biobank.
Expression of 649 tumor-enriched differentially expressed genes selected by an elastic net regularized Cox proportional hazards model of overall survival was weighted by hazard ratios. Clustering was performed with ConsensusClusterPlus, retaining clusters with high consensus. We identified five clusters, with three neuroendocrine positive (NE+). Group overall survival (OS) differed significantly with log-rank p-value ≈ 0.010. NE+ clusters 1 and 3 were composed mostly of stage 1A patients with mOS <2 years and >6 years, respectively.