一种超高灵敏度肿瘤指导MRD检测的分析验证

Circulating tumor DNA (ctDNA) is a biomarker for molecular residual disease (MRD), i.e., tumor cells that persist during or after cancer treatment. ctDNA can be distinguished from normal cell-free DNA (cfDNA) based on the presence of tumor-specific somatic variants. First-generation tumor-informed MRD assays based on exome sequencing typically interrogate a small number of tumor-specific variants, limiting their sensitivity. We have developed a second-generation, tumor-informed MRD assay using matched formalin fixed paraffin embedded (FFPE) tumor tissue, including core needle biopsies, and normal whole genome sequencing to power expansive variant discovery and selection of panels consisting of up to 1000 targets optimized for assay sensitivity. Here we validate this assay using more than 100 unique cancer-affected clinical sample sets, more than 30 unique healthy donors and 3 unique sets of matched tumor and normal cell lines. Panels were generated from paired tumor and normal sample sets using as little as 10 ng of DNA, with highly reproducible performance across replicate panel designs. More than 4400 plasma and contrived sample tests were run across a range of assay conditions, including cfDNA input from 7.5 to 40 ng and panel sizes ranging from 500 to 1000 targets. Contrived samples were tested at tumor DNA concentrations ranging from 0.78 to 1,000,000 parts per million (ppm). We found the Limit of Detection with 95% sensitivity (LoD 95 ) to be <5 ppm in units of genome equivalents (comparable to <2.5 ppm in units of variant allele frequency) under typical assay conditions, and that LoD 95 scales with cfDNA input amount and panel size. We determined the linear range of the assay to be from 1.56 to 1,000,000 ppm, with high quantitative precision down to 18 ppm. Across all tested indications, including breast, renal, colorectal, ovarian, endometrial and lung cancers from stages II-IV, clinical sensitivity is >85% and clinical specificity is >99%, where undetected samples were exclusively in low shedding indications, such as renal and HR+/HER2- breast cancer. Clinical sensitivity is 100% with clinical specificity >99% in colorectal, ovarian, endometrial and lung cancers. In conclusion, we have developed and validated a quantitative and ultra-high sensitivity tumor-informed MRD assay suitable for challenging, low shedding applications such as breast and renal cancer or post-treatment monitoring.