Accurate detection of DNA variants and ecDNA from prostate FFPE biopsies using LinkPrep

Detecting genomic variants and structural rearrangements in formalin-fixed paraffin-embedded (FFPE) tumor biopsies simultaneously remains challenging and lacks comprehensive clinical utility assessments. Here we performed the novel Dovetail®-FFPE (LinkPrep), an in vitro transposition Hi-C assay free from restriction enzyme bias to 20 prostate cancer samples. These included 5 normal adjacent to tumor (benign), 10 primary localized, and 10 metastatic castration-resistant prostate cancer (mCRPC) patient samples. By leveraging patient and site-specific in-house matched fresh-frozen (FF) whole-genome sequencing (WGS), exome capture sequencing, transcriptomics sequencing, and a publicly available cohort of 80 mCRPC FF Hi-C samples data, we performed a comprehensive evaluation on FFPE LinkPrep library preparation and sequencing performance. Our evaluation covered multiple aspects including 3D genome contact detection, single nucleotide variant (SNV)/indel calling, copy number variation analysis, structural variant (SV) detection, ecDNA detection and compartments definition.Our analysis shows that FFPE sample achieves good contact capture efficiency that is slightly lower but comparable performance (Wilcoxon test p=0.33) to the FF Hi-C data for mCRPC tumors. For SNV/INDEL calling, using FF WGS or exome capture sequencing as the gold standard, the assay reached a mean recall rate of 97% across five samples based on two different variant- calling algorithms. By comparing with 2 FF WGS, 4 transcriptomic and 80 external Hi-C data, we further demonstrated that FFPE samples can accurately identifies whole-genome copy number variations and chromatin compartments. Additionally, we developed a robust SV-calling pipeline with carefully designed denoising steps suitable for FFPE samples that accurately detected common structural variants in prostate cancer, such as TMPRSS2-ERG fusions (6/7) and FOXA1mutations (2/2). Lastly, by applying the new developed Hi-C specific pipeline for extrachromosomal DNA (ecDNA) detection, we identified AR ecDNA in one sample, which was supported by matched FF WGS AmpliconSuite result. Overall, the comprehensive analysis and performance evaluation demonstrates that FFPE LinkPrep accurately captures DNA-level variants and holds strong potential for application in FFPE clinical samples.