Morphomechanically-Informed Spatial Curvature Sequencing in Prostate Cancer

Area of Science:

• Urology
• Computational Biology
• Cancer Research

Background:

• Prostate cancer diagnosis relies on Gleason grading of gland morphology.
• Molecular drivers of prostate gland shape changes are not well understood.
• Spatial omics data offers potential to link morphology and molecular state.

Purpose of the Study:

• Introduce CurvSeq and CurvSee, novel frameworks for analyzing spatial sequencing and imaging data.
• Integrate prostate gland boundary curvature, pocket architecture, and microenvironment with molecular profiles.
• Investigate morphomechanical relationships in prostate adenocarcinoma and identify cancer-associated features.

Main Methods:

• Developed CurvSeq for spatial transcriptomics and CurvSee for proteomics/imaging.
• Segmented individual prostate glands and extracted contour features like curvature and concavities.
• Projected morphomechanical features onto spatially resolved gene and protein expression data across five independent datasets.

Main Results:

• CurvSeq differentiated benign and Gleason Grade 1 (GG1) glands, identifying cancer markers (PCA3, AMACR) in GG1 and specific programs in benign glands.
• A morphomechanical score in Visium data correlated gland shape, microenvironment, and remodeling genes (MMP7).
• CurvSeq identified neuroendocrine-like regions in GG4 glands associated with MMP7, COL1A1, and immune cells; CurvSee distinguished Gleason states using morphology and protein data.

Conclusions:

• CurvSeq and CurvSee provide a quantitative framework linking prostate gland architecture, microenvironment, and molecular state.
• Prostate gland morphology is a valuable feature that can be integrated with spatial omics.
• These methods identify morphomechanical niches critical for understanding prostate cancer progression.