Background: Identification of actionable mutations in circulating tumor DNA (ctDNA) enables gene-targeted therapy of solid tumors based on a simple blood test. NGS methods that attach a unique molecular identifier (UMI) to each DNA molecule reduce variant allele fraction (VAF) limit of detection (LOD) to <0.01% at a cost of 10-20-fold higher sequencing requirement. Importantly, the VAF lower limit of detection (LOD) for analysis of ctDNA specimens does not typically extend below 0.5% due to limits of ctDNA specimen quantity. For example, the average ctDNA specimen is 32 ng (10,000 genome copies) and at 0.1% VAF, 10 mutant molecules would be added to library prep. At this level, signal loss during library prep (typically 70-90%) will be associated with stochastic sampling variation. For example, in the Sequencing Quality Control Consortium (SEQC) 2 study, we measured inter-site and inter-replicate reproducibility of VAF measurement in 50 ng (15,000 genome copies) of synthetic ctDNA reference material. Thus, even using UMI NGS, measurement, reproducibility at VAF >0.5% was >95%, but reproducibility at 0.1-0.5% VAF was about 70%. Given limited value of UMI NGS when size of ctDNA specimen restricted VAF LOD to 0.5%, we tested the hypothesis that synthetic internal standard (IS) spike-in molecules provide reliable alternative quality control while eliminating UMI-imposed sequencing burden.

Methods: We synthesized IS for 61 actionable mutations, cloned them into pUC vectors, confirmed each IS to be wild-type sequence, linearized, quantified abundance and combined at a 1:1 genome copy mixture. An aliquot of IS mixture was added to the ctDNA reference material at 1:1 genome copy prior to library prep, followed by Illumina HiSeq sequencing, separation of target from IS reads using custom splitter, then pipeline analysis on Qiagen CLC Genomics Workbench.

Results: Analysis of each target relative to IS reliably identified base-substitution sequencing errors at each measured actionable mutation site and determined that VAF for each was <0.2%. Further, although a typical UMI method requires >200,000 reads per target in each sample, with analysis relative to IS spike-ins, 20,000 reads was sufficient (10,000 reads for the specimen and 10,000 reads for the spike-in). Because the IS spike-in molecules were validated to be wild-type, any variants were assumed to be due to technical error. Thus, a z-score comparing sequence count for a) specimen wild-type, b) specimen mutant, c) IS wild-type, and d) IS mutant enabled calculation of confidence level regarding each biological variant call.

Conclusion: Preliminary data indicate that IS spike-in mixtures enable reliable analysis of ctDNA mutation fraction to LOD of 0.5% without need for UMI.

Disclaimer: The views expressed here are those of the authors only and do not necessarily express the views/policies of the FDA.

Citation Format: Daniel J. Craig, Erin L. Crawford, Joshua Xu, Thomas M. Blomquist, Leihong Wu, Thomas Morrison, James C. Willey. Novel method for NGS analysis of actionable mutations in circulating tumor DNA specimens: improved quality control and 20-fold lower sequencing required [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 432.