The analysis of circulating cell-free DNA (cfDNA) allows noninvasive monitoring of cancer genomes. The estimation of tumor-derived mutant allele fractions (MAF) in cfDNA can guide the selection of appropriate molecular analysis and may also reveal clinically relevant information. In this study, we evaluate two approaches to estimate tumor-derive MAF: shallow whole genome sequencing and fragment size analysis.


We performed WGS at a shallow depth (~0.1x) on cfDNA samples of 118 metastatic adult cancer patients and 21 advanced pediatric cancer patients. We used genome-wide Z-scores, a metric to determine tumor-derived MAF, and the ratio of short:long (100 - 150 bp : 163 - 169 bp) cfDNA fragments to predict mutant allele fraction determined by MSK-IMPACT on the same cfDNA samples.


We found that genome wide z-score derived from sWGS separates the size distributions into distinct regions where high allele fractions are associated with a shift in the distributions to shorter cfDNA fragments. In addition, we incorporated a size ratio of short to long fragments from sWGS data to predict median mutant allele fraction, we obtained a threshold of 1.25, an AUC of 0.81, a PPV of 79% (27/34), and an NPV of 75% (79/ 105). Combining both Z-score and size ratio, we obtained an AUC of 0.94, a PPV of 85% (29/34), and an NPV of 93% (98/105). From 118 patients, we found that patients with high median mutant allele fraction (> 5%) were associated with shorter overall survival after the time of blood draw. Patients with high genome-wide z-score (> 2.5) were associated with poor survival rates after the time of blood draw with a p-value of (4.87 x 10-4). We saw the same association between cfDNA MAF (> 5%) and overall survival (p-value 6.82 x 10-4), suggesting that sWGS-derived genome-wide z-score is an accurate estimate of tumor mutant allele fractions in cfDNA, both are associated with patients' overall survival.


It is feasible to infer tumor-derived MAF from plasma cfDNA using genome-wide z-score estimated from sWGS analysis, or by studying the ratio of short:long fragments from the size profiles. Such information can guide the selection of appropriate molecular analysis strategy and reveal clinically relevant information.

Citation Format: Julie Yang, Francesco Marass, Peter Ulz, Maha Shady, Michael L. Cheng, Prachi Kothari, Erika Gedvilaite, Saira Somnay, Neerav Shukla, Shakeel Modak, Emily K. Slotkin, Dana E. Rathkopf, Iyer Gopaumar, Howard I. Scher, Pedram Razavi, Darren R. Feldman, Bob T. Li, Paul B. Chapman, Jonathan E. Rosenberg, Dean F. Bajorin, Howard I. Scher, Michael F. Berger, Venkatraman E. Seshan, Nicholas D. Socci, David B. Solit, Ellen Heitzer, Dana W. Y. Tsui. Estimation of tumor-derived mutant allele fractions in cfDNA by shallow whole genome sequencing and fragment size analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr LB-227.