PO.BCS01.12 · 生物信息与计算
gOS：一个应用于修复后FFPE、血液全基因组及临床靶向检测的整合分子肿瘤学解读框架
gOS: A total molecular oncology interpretation framework applied to repaired FFPE and heme whole genome and clinical targeted assays
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
癌症分子诊断在多种检测和互不衔接的软件之间仍然碎片化，使分子病理学家在肿瘤科医生能够依据建议采取行动之前，不得不手动关联QC、纯度/倍性和变异类别。我们通过两项互补的全基因组WGS检测——一项修复后FFPE实体瘤检测和一项仅用血液的血液肿瘤检测——来解决这一问题，将SNV/indel、CNA、SV/融合、LOH及特征谱整合到具有类似panel性能的单一DNA工作流程中。与此同时，我们构建了整合基因组肿瘤学系统（gOS），它将WGS、靶向panel、甲基化和RNA融合整合到一个连贯的浏览器中，具备变异分级、临床验证和面向发现的探索功能。
在NYU，标准诊疗包括NYU-PACT（607基因DNA panel）、用于实体瘤的Illumina MethylationEPIC v2.0，以及用于驱动性血液肿瘤融合的Oncomine（40基因DNA）加Myeloseqer。为试点用WGS替代DNA/RNA panel，我们采用了酶切法（Watchmaker Genomics）和FFPE损伤修复步骤，以改善覆盖均一性并降低伪影。我们分析了23例配对正常样本的混合实体瘤和22例血液系统恶性肿瘤；其中7例实体FFPE还进行了未修复处理以作对比。gOS通过整合的Nextflow流程处理了全部55例病例，并提供了交互式的WGS与SoC比较。
未修复的FFPE表现出富AT区域丢失和兆碱基级别的覆盖偏倚；修复恢复了均一性并抑制了伪影。在一例BRCA1突变/LOH病例中，未修复WGS掩盖了HRD（HRDetect 0.004；B1+2 0.2），但经修复后得以恢复（HRDetect 0.99；B1+2 0.78）。在23例修复后FFPE实体瘤中，WGS相较NYU-PACT达到96%的灵敏度（298/310个热点）和接近100%的特异性（6,740个SoC野生型位点中仅1个）。血液肿瘤WGS与Myeloseqer融合和Oncomine突变达到100%一致性。
综上，互补的WGS检测与gOS共同提供了类似panel的准确性，同时具有更广泛的泛变异覆盖和单一、连贯的解读视图，支持将WGS作为碎片化DNA/RNA panel的实用替代方案——尤其是在历来具有挑战性的FFPE组织中——并通过统一的多组学解读实现临床效用的拓展。
查看英文原文 English abstract
Cancer molecular diagnostics remain fragmented across multiple assays and disjointed software, leaving molecular pathologists to manually link QC, purity/ploidy, and variant classes before oncologists can act on recommendations. We address this with two complementary, genome-wide WGS assays-a repaired-FFPE solid tumor test and a blood-only heme test-that consolidate SNVs/indels, CNAs, SVs/fusions, LOH, and signatures into a single DNA workflow with panel-like performance. In parallel, we built the total genome Oncology System (gOS), which integrates WGS, targeted panels, methylation, and RNA fusions into a coherent browser with variant tiering, clinical validation, and discovery-oriented exploration.
At NYU, standard care comprises NYU-PACT (607-gene DNA panel), Illumina MethylationEPIC v2.0 for solid tumors, and Oncomine (40-gene DNA) plus Myeloseqer for driver heme fusions. To pilot a WGS drop-in for DNA/RNA panels, we used enzymatic shearing (Watchmaker Genomics) and a FFPE damage-repair step that improves coverage uniformity and lowers artifacts. We analyzed 23 mixed solid tumors with paired normals and 22 hematologic malignancies; seven solid FFPE were also prepped without repair for comparison. gOS processed all 55 cases via an integrated Nextflow pipeline and provided interactive WGS vs SoC comparison.
Unrepaired FFPE showed AT-rich dropout and megabase-scale coverage bias; repair restored uniformity and suppressed artifacts. In a BRCA1-mutant/LOH case, HRD was obscured in unrepaired WGS (HRDetect 0.004; B1+2 0.2) but recovered with repair (HRDetect 0.99; B1+2 0.78). Across 23 repaired FFPE solids, WGS vs NYU-PACT achieved 96% sensitivity (298/310 hotspots) and near-100% specificity (1/6,740 SoC-wild-type positions). Heme WGS showed 100% concordance with Myeloseqer fusions and Oncomine mutations.
Together, the complementary WGS assays and gOS deliver panel-like accuracy with broader, pan-variant coverage and a single, coherent interpretive view, supporting WGS as a practical replacement for fragmented DNA/RNA panels-especially in historically challenging FFPE tissue-and enabling expanded clinical utility through unified multi-omic interpretation.
利益披露 Disclosure
K. Hadi,
Isabl, Inc. Employment.
A. Deshpande,
Isabl, Inc. Employment.
S. Dider, None..
C. Xanthopoulakis, None..
J. Rafailov, None..
M. Imielinski, None.