开发全面的肿瘤知情ctDNA工作流程，利用多样化的肿瘤图谱输入实现超敏分子残留疾病检测

Background: Circulating tumor DNA (ctDNA) analysis represents a transformative approach to molecular residual disease (MRD) surveillance, enabling non-invasive disease monitoring and treatment response assessment. While tumor-agnostic methodologies offer broad applicability, tumor-informed strategies leveraging patient-specific genomic alterations can provide enhanced sensitivity for detecting low-burden disease. We developed an integrated wet lab and bioinformatics workflow capable of identifying and monitoring structural variants (SVs), single nucleotide variants (SNVs), and insertions/deletions (Indels) across multiple genomic profiling platforms. Methods: Patient-specific variant discovery was performed using whole genome sequencing (WGS), comprehensive genomic profiling (CGP), or whole exome sequencing (WES) using formalin-fixed paraffin-embedded tissue (FFPET) specimens. Our proprietary informatics pipeline annotated, filtered, and prioritized variants based on stringent quality metrics and variant allele frequency. To evaluate WGS inputs, two clinical metastatic castration-resistant prostate cancer (mCRPC) specimens with matched FFPET and plasma were used to design multiplexed droplet digital PCR (ddPCR) assays for detection of up to 12 SVs per reaction well. For cases with CGP inputs, off-the-shelf duplex ddPCR assays were used to monitor 6 unique SNVs detected across clinical mCRPC patients. The informatics pipeline was also employed to identify SVs, SNVs, and small Indels from patients with WES data from tissue. Results: WGS libraries were successfully generated and passed quality control using as little as 40 ng FFPET DNA. Sequencing identified multiple high-confidence SVs, from which bespoke assays were designed. Assays were pooled and evaluated for cross-reactivity with wild-type genomic DNA and no-template controls. All CGP-derived targets were successfully detected in matched baseline plasma specimens with variant allele frequencies as low as 0.12 percent. WES data analysis from 8 subjects identified multiple SVs, SNVs, and Indels using a proprietary informatics pipeline. However, no WES-identified SVs were suitable for ddPCR assay design, suggesting SNV and Indel targets may be more appropriate for this data input type. Conclusions: This study demonstrates successful development of a comprehensive ctDNA workflow for MRD detection using mCRPC specimens. The workflow accommodates multiple sequencing input types (WGS, CGP, WES) and successfully detects low-frequency variants in plasma using multiplex ddPCR. These data suggest that WGS-derived SVs and CGP- or WES-derived SNVs/Indels represent complementary strategies for tumor-informed MRD monitoring. Further validation studies with larger cohorts from various cancer types are planned.