Evaluating the analytical validity of circulating tumor DNA sequencing assays for precision oncology

Ira W. Deveson,Binsheng Gong,Kevin Lai,Jennifer S. LoCoco,Todd A. Richmond,Jeoffrey Schageman,Zhihong Zhang,Natalia Novoradovskaya,James C. Willey,Wendell Jones,Rebecca Kusko,Guangchun Chen,Bindu Swapna Madala,James Blackburn,Igor Stevanovski,Ambica Bhandari,Devin Close,Jeffrey Conroy,Michael Hubank,Narasimha Marella,Piotr A. Mieczkowski,Fujun Qiu,Robert Sebra,Daniel Stetson,Lihyun Sun,Philippe Szankasi,Haowen Tan,Lin-ya Tang,Hanane Arib,Hunter Best,Blake Burgher,Pierre R. Bushel,Fergal Casey,Simon Cawley,Chia-Jung Chang,Jonathan Choi,Jorge Dinis,Daniel Duncan,Agda Karina Eterovic,Liang Feng,Abhisek Ghosal,Kristina Giorda,Sean Glenn,Scott Happe,Nathan Haseley,Kyle Horvath,Li-Yuan Hung,Mirna Jarosz,Garima Kushwaha,Dan Li,Quan-Zhen Li,Zhiguang Li,Liang-Chun Liu,Zhichao Liu,Charles Ma,Christopher E. Mason,Dalila B. Megherbi,Tom Morrison,Carlos Pabón-Peña,Mehdi Pirooznia,Paula Z. Proszek,Amelia Raymond,Paul Rindler,Rebecca Ringler,Andreas Scherer,Rita Shaknovich,Tieliu Shi,Melissa Smith,Ping Song,Maya Strahl,Venkat J. Thodima,Nikola Tom,Suman Verma,Jiashi Wang,Leihong Wu,Wenzhong Xiao,Chang Xu,Mary Yang,Guangliang Zhang,Sa Zhang,Yilin Zhang,Leming Shi,Weida Tong,Donald J. Johann,Timothy R. Mercer,Joshua Xu,
DOI: https://doi.org/10.1038/s41587-021-00857-z
IF: 46.9
2021-04-12
Nature Biotechnology
Abstract:Circulating tumor DNA (ctDNA) sequencing is being rapidly adopted in precision oncology, but the accuracy, sensitivity and reproducibility of ctDNA assays is poorly understood. Here we report the findings of a multi-site, cross-platform evaluation of the analytical performance of five industry-leading ctDNA assays. We evaluated each stage of the ctDNA sequencing workflow with simulations, synthetic DNA spike-in experiments and proficiency testing on standardized, cell-line-derived reference samples. Above 0.5% variant allele frequency, ctDNA mutations were detected with high sensitivity, precision and reproducibility by all five assays, whereas, below this limit, detection became unreliable and varied widely between assays, especially when input material was limited. Missed mutations (false negatives) were more common than erroneous candidates (false positives), indicating that the reliable sampling of rare ctDNA fragments is the key challenge for ctDNA assays. This comprehensive evaluation of the analytical performance of ctDNA assays serves to inform best practice guidelines and provides a resource for precision oncology.
biotechnology & applied microbiology
What problem does this paper attempt to address?
The paper is primarily dedicated to addressing the issue of analytical effectiveness of circulating tumor DNA (ctDNA) sequencing in precision oncology. Specifically, the article focuses on the following aspects: 1. **Detection Sensitivity and Accuracy**: The paper evaluates the performance of five industry-leading ctDNA sequencing methods at different variant allele frequencies (VAF). When the variant allele frequency is above 0.5%, all methods can detect ctDNA mutations with high sensitivity, accuracy, and reproducibility; however, at frequencies below 0.5%, detection becomes unreliable, and the performance varies significantly between different methods. 2. **Technical Challenges**: Detecting rare somatic mutations from limited input material is highly challenging due to the low concentration and small fragment size of cell-free DNA. Additionally, ctDNA sequencing is influenced by various experimental variables and artifacts, such as PCR amplification, target region enrichment, and sequence alignment issues. 3. **Coverage Depth and Heterogeneity**: Coverage depth is a key variable in ctDNA sequencing, and high coverage depth is crucial for the sensitive detection of low-frequency mutations. Besides depth, uniform coverage within the target region is also important to ensure high sensitivity and reproducibility. 4. **Impact of Input DNA Amount**: Increasing the DNA input amount generally improves fragment depth, sensitivity, and reproducibility. However, the availability of cell-free DNA in clinical samples is limited, necessitating improved extraction efficiency or increased patient blood collection to augment input material. 5. **Inter-Laboratory Variability**: The methods tested showed good robustness across different laboratories, mainly affected by random rather than systematic variations. This indicates that the robustness of ctDNA sequencing methods concerning technical variables is crucial for clinical applications. Through these evaluations, the paper aims to provide a basis for best practice guidelines for ctDNA sequencing and to offer resources for the application of precision oncology.