3D open-top light-sheet microscopy and 3D microdissection of neuoadjuvant-treated primary prostate cancer reveals latent subclonal mutations.

Abstract

e17109 Background: Intratumoral molecular heterogeneity is a well described phenomenon in primary prostate cancer. Increased genomic heterogeneity is associated with more aggressive clinical behavior, including recurrence and metastasis. However, current methods to characterize heterogeneity suffer from tradeoffs between generating high quality morphologic information, correlating morphology with genomic alterations, and workflow considerations. We describe a technology to image prostate samples in 3D, perform 3D microdissection of regions of interest (ROIs), and preserve nucleic acids for DNA sequencing. Methods: Samples (N = 18) were obtained from patients (N = 12) enrolled in an open-label, single-site, Phase II neoadjuvant trial in men with NCCN high to very-high risk PC. Patients received 12 weeks of apalutamide, abiraterone acetate plus prednisone, degarelix, and indomethacin followed by radical prostatectomy (RP). Fresh frozen prostate samples (~2x2x0.4cm) were obtained from RP specimens. Frozen samples were fixed in an ethanol-based fixative, stained with nuclear (TOPRO-3) and general protein (eosin) fluorescent dyes, cleared with ethyl cinnamate, and imaged using 3D open-top light-sheet (OTLS) fluorescence microscopy. Image-guided 3D microdissection was performed on ROIs with distinct morphologies. Microdissected ROIs measured 0.5x0.5x2mm. Microdissected samples were sent for DNA extraction and sequencing using the comprehensive clinically-validated UW-OncoPlex NGS assay. Results: The tissue from neoadjuvant-treated patients contained sparse carcinoma, with cancer volumes ranging from 0.01-0.3 cm2 and cellularity ranging from 2%-30%. Molecular alterations were detected in a sample with only 0.07 cm2 volume of carcinoma with 5% cellularity. The read depth of 3D microdissected samples was 3.9 times greater than the read depth of FFPE samples from the same patients. At least two morphologically distinct ROIs were 3D microdissected from 11 of 12 patients. Four of the 18 samples had molecular alterations that differed between ROIs, i.e. intra-sample molecular heterogeneity. For example, one sample with three microdissected ROIs had a PTEN alteration exclusive to a single ROI. Conclusions: The findings show the potential for 3D microdissection of morphologically distinct subclones in samples from neoadjuvant-treated patient, to enable sequencing of specific cellular populations. The sparsity of carcinoma necessitated a spatially targeted approach for extracting material for sequencing. The increased yield from avoiding FFPE processing and carcinoma enrichment from targeted 3D microdissection of carcinoma ROIs enabled the detection of spatially distinct molecular subclones. Future work will include the analysis of additional samples and implementation of AI-guided ROI detection to streamline workflows.