Analytical and clinical validation of a personalized, tumor-informed circulating tumor DNA assay for minimal residual disease detection.

e14545 Background: The detection of minimal residual disease (MRD) through circulating tumor DNA (ctDNA) analysis provides an advanced molecular method for early identification of surgical patients at high risk of recurrence. In this study, we present the development of mClear, a personalized, tumor-informed next-generation sequencing (NGS) assay for MRD detection. Furthermore, we report the analytical and clinical validation processes conducted to evaluate the performance of the MRD assay. Methods: mClear is designed to assess MRD in patients with solid tumors. Whole exome sequencing (WES) was performed on tumor tissues and matched peripheral blood samples to detect somatic variants comprehensively. Personalized MRD assays were carefully designed by selecting up to 40 prioritized variants, including single nucleotide variants (SNVs), small insertions and deletions (InDels), and fusions while ensuring coverage of 21 additional key driver genes. For analytical validation, we utilized a pan-cancer MRD ctDNA quality control product (LDT-999, LDT Bioscience), characterized by known variant allele frequency (VAF) gradients of 0%, 0.005%, 0.05%, and 0.5%, confirmed through droplet digital PCR (ddPCR). We conducted 18 replicate tests at VAF gradients of 0.5% and 0.05%, 23 replicates at the 0.005% VAF gradient, and 16 replicates at the 0% VAF gradient. For clinical validation, preoperative and postoperative blood samples from a cohort of 264 pan-cancer patients were analyzed. ctDNA collected for MRD detection underwent ultra-deep next-generation sequencing (100,000× average depth of coverage). The library preparation process commenced with an input of 30ng DNA. Results: At VAF gradients of 0.5% and 0.05%, the assay achieved an excellent sensitivity of 100%, and at VAF gradients of 0.005% and 0%, it demonstrated a sensitivity of 95.7% and a specificity of 100%. Preoperative ctDNA was detected in 86.7% (72/83) of the pan-cancer patients with 71.4% (5/7), 90.9% (10/11), 85.5% (47/55), and 100.0% (10/10) in pathological stage (pStage)I, pStage II, pStage III, and pStage IV, respectively. In the pan-cancer cohort, the positive rate of longitudinal MRD was 35.7% (75/210) with 17.8% (21/118), 50.0% (13/26), 55.1% (27/49), and 82.4% (14/17) in pStage I, II, III, and IV, respectively. In four patients who relapsed, MRD detection during surveillance preceded radiological recurrence by a lead time of 1-8 months. Conclusions: Our personalized MRD assay demonstrated excellent analytical sensitivity and specificity with a limit of detection (LoD) of 0.005% VAF at the sample level. Furthermore, it displayed high clinical sensitivity for the detection of ctDNA in the pan-cancer cohort, highlighting its potential for monitoring of recurrence.