Dissecting molecular, pathological, and clinical features associated with tumor neural/neuroendocrine heterogeneity

Abstract

Summary: Lineage plasticity, especially transdifferentiation between neural/neuroendocrine (NE) and non-NE lineage, has been observed in multiple cancer types and linked to increased tumor aggressiveness. However, existing NE/non-NE subtype classifications in various cancer types were established through ad hoc approaches in different studies, making it difficult to align findings across cancer types and extend investigations to new datasets. To address this issue, we developed a generalized strategy to generate quantitative NE scores and a web application to facilitate its implementation. We applied this method to nine datasets covering seven cancer types, including two neural cancers, two neuroendocrine cancers, and three non-NE cancers. Our analysis revealed significant NE inter-tumoral heterogeneity and identified strong associations between NE scores and molecular, histological, and clinical features, including prognosis in different cancer types. These results support the translational utility of NE scores. Overall, our work demonstrated a broadly applicable strategy for determining the NE properties of tumors.

Keywords

• Computational bioinformatics
• Cancer systems biology
• Cancer