Genome-wide cell-free DNA mutational integration enables ultra-sensitive cancer monitoring

Zviran, Asaf; Schulman, Rafael C.; Shah, Minita; Hill, Steven T.; Deochand, Sunil; Khamnei, Cole C.; Maloney, Dillon; Patel, Kristofer; Liao, Will; Widman, Adam; Wong, Phillip; Callahan, Margaret K.; Ha, Gavin; Reed, Sarah C.; Rotem, Denisse; Frederick, Dennie T.; Sharova, Tatyana; Miao, Benchun; Kim, Tommy; Gydush, Greg; Rhoades, Justin; Huang, Kevin; Omans, Nathaniel D.; Bolan, Patrick O.; Lipsky, Andrew; Ang, Chelston; Malbari, Murtaza; Spinelli, Catherine F.; Kazancioglu, Selena; Runnels, Alexi; Fennessey, Samantha; Stolte, Christian; Gaiti, Federico; Inghirami, Giorgio; Adalsteinsson, Viktor A.; Houck-Loomis, Brian; Ishii, Jennifer; Wolchok, Jedd D.; Boland, Genevieve M.; Robine, Nicolas; Altorki, Nasser K.; Landau, Dan A.

doi:10.1038/s41591-020-0915-3

Cited by 256 publications

(306 citation statements)

References 48 publications

Supporting

Mentioning

267

Contrasting

Order By: Relevance

“…This is significant because (i) detecting MRD remains a significant unmet medical need, and (ii) while MRD detection correlates with the number of tumor mutations tracked in cfDNA 27,34,35 , existing techniques have had limited breadth or depth. For instance, cancer gene panels typically cover just a few mutations per patient 37 ; patient-specific assays track tens to hundreds 27,33 ; and whole-genome sequencing remains far too costly to apply beyond minimal depth 46 . Using MAESTRO, we found many more mutations detected at limiting dilutions such as 1/100k, from about 5 when 438 were tracked to almost 200 when 10,000 were tracked.…”

Section: Discussionmentioning

confidence: 99%

“…We also focused on enrichment of point mutations, but expect that MAESTRO could also be useful for tracking other types of alterations such as insertions and deletions or structural variants. While tracking more mutations per patient could increase the number of unique cfDNA molecules sampled (and therefore, the detection limit for MRD) 27,35,37,46 , it will never be possible to detect MRD at tumor fractions below sequencing error rates. Accordingly, we opted to employ the most accurate sequencing method, duplex sequencing.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

MAESTRO affords ‘breadth and depth’ for mutation testing

Gydush

Nguyen

Bae

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

The ability to assay large numbers of low-abundance mutations is crucial in biomedicine. Yet, the technical hurdles of sequencing multiple mutations at extremely high depth and accuracy remain daunting. For sequencing low-level mutations, it’s either ‘depth or breadth’ but not both. Here, we report a simple and powerful approach to accurately track thousands of distinct mutations with minimal reads. Our technique called MAESTRO (minor allele enriched sequencing through recognition oligonucleotides) employs massively-parallel mutation enrichment to empower duplex sequencing—one of the most accurate methods—to track up to 10,000 low-frequency mutations with up to 100-fold less sequencing. In example use cases, we show that MAESTRO could enable mutation validation from cancer genome sequencing studies. We also show that it could track thousands of mutations from a patient’s tumor in cell-free DNA, which may improve detection of minimal residual disease from liquid biopsies. In all, MAESTRO improves the breadth, depth, accuracy, and efficiency of mutation testing.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

MAESTRO affords ‘breadth and depth’ for mutation testing

Gydush

Nguyen

Bae

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Integration of additional blood-based biomarkers (e.g., blood-based tumor mutational burden, immune cell proportions) with ctDNA kinetics may further improve the accuracy of immunotherapy response prediction (18). Other technologies that have demonstrated potential relevance in the MRD setting include whole-genome sequencing of ctDNA based on the cumulative signals from thousands of somatic mutations harbored by many solid tumors (19). It is expected that over time, an increasing number of interception clinical trials will be conducted, investigating new drugs or drug combinations that have demonstrated an adequate safety profile as well as established evidence of antitumor activity in the recurrent or metastatic setting.…”

Section: Cancer Interception Trials For Molecular Residual Diseasementioning

confidence: 99%

The Future of Clinical Trial Design in Oncology

et al. 2021

View full text Add to dashboard Cite

Clinical trials represent a fulcrum for oncology drug discovery and development to bring safe and effective medicines to patients in a timely manner. Clinical trials have shifted from traditional studies evaluating cytotoxic chemotherapy in largely histology-based populations to become adaptively designed and biomarker-driven evaluations of molecularly targeted agents and immune therapies in selected patient subsets. This review will discuss the scientific, methodological, practical, and patient-focused considerations to transform clinical trials. A call to action is proposed to establish the framework for next-generation clinical trials that strikes an optimal balance of operational efficiency, scientific impact, and value to patients.Research.

show abstract

“…However, adding genome-wide information could improve the classification. According to a recent study on genome-wide liquid biopsies of postoperative early stage residual cancers, the integration of genome-wide mutation data allowed sensitive residual disease detection by overcoming the limitations of sparsity [17]. Additionally, previous work on WGS data from solid tissue biopsies have already established that somatic SNV density at 1 Mb scale is the most prominent predictor of cancer type as it represents the genomic imprint of the cell of origin chromatin organization, with passenger somatic SNVs being the most prominent contributors [4, 5].…”

Section: Introductionmentioning

confidence: 99%

“…In our work, we explore the utilization of sparse genome-wide somatic mutation data in the classification of the cell of origin of cancer. High quality genome-wide somatic mutation data obtained from ctDNA is very scarce [17], and by no means sufficient to support the training of robust classifiers. Therefore sparse SNV samples are generated based on WGS of primary cancer samples from the PCAWG dataset [18] to model ctDNA conditions.…”

Section: Introductionmentioning

confidence: 99%

Cancer type classification in liquid biopsies based on sparse mutational profiles enabled through data augmentation and integration

Danyi

Jager

Ridder

2021

Preprint

View full text Add to dashboard Cite

Identifying the cell of origin of cancer is important to guide treatment decisions. However, in patients with 'cancer of unknown primary' (CUP), standard diagnostic tools often fail to identify the primary tumor. As an alternative, machine learning approaches have been proposed to classify the cell of origin based on somatic mutation profiles in the genome of solid tissue biopsies. However, solid biopsies can cause complications and certain tumors are not accessible. A promising alternative would be liquid biopsies, which contain ctDNA originating from the tumor. Problematically, somatic mutation profiles of tumors obtained from liquid biopsies are inherently extremely sparse and current machine learning models fail to perform in this setting. Here we propose an improved machine learning method to deal with the sparse nature of liquid biopsy data. Firstly, we downsample the SNVs in the samples in order to mimic sparse data conditions. Then extensive data augmentation is performed to artificially increase the number of training samples in order to enhance model robustness under sparse data conditions. Finally, we employ data integration to merge information from i) somatic single nucleotide variant (SNV) density across the genome, ii) somatic SNVs in driver genes and iii) trinucleotide motifs. Our adapted method achieves an average accuracy of 0.88 on the data where only 70% of SNVs are retained, which is comparable to an average accuracy of 0.87 with the original model on the full SNV data. Even when only 2% of the data is retained, the average accuracy is 0.65 compared to 0.41 with the original model. The method and results presented here open the way for application of machine learning in the detection of the cell of origin of cancer from sparse liquid biopsy data.

show abstract

Genome-wide cell-free DNA mutational integration enables ultra-sensitive cancer monitoring

Cited by 256 publications

References 48 publications

MAESTRO affords ‘breadth and depth’ for mutation testing

MAESTRO affords ‘breadth and depth’ for mutation testing

The Future of Clinical Trial Design in Oncology

Cancer type classification in liquid biopsies based on sparse mutational profiles enabled through data augmentation and integration

Contact Info

Product

Resources

About