Thread by @landau_lab

Excited to share our work @NatureMedicine offering a novel framework [MRDetect] for ultra-sensitive cancer monitoring through whole genome mutational integration of cfDNA go.nature.com/2XmxJZv
1/
#LiquidBiopsy #WGS #cfDNA #MachineLearning
👇👇👇

Therapy monitoring is a central pillar of modern medicine. And yet, in many areas of oncology, we lack sensitive monitoring tools for residual disease detection. ctDNA carries the potential to change this, but is scarce in low-burden disease.

The central paradigm in the field aims to overcome ctDNA scarcity through deep targeted sequencing of informative sites. However, we show that limited number of fragments imposes a hard ceiling on cfDNA deep targeted sequencing, with sensitivity limited ~1:1,000 per site.

We propose that genome-wide integration, across many thousands of mutations, can radically change the detection landscape.

This required rethinking the mutation detection process, given that only a single supporting mutated read is expected per locus, at best. We developed a read-centric (instead of locus-centric) caller, enabling machine learning on millions of reads for 10-100X noise reduction

Through mixing studies, we show that this perspective enables detection of tumor fractions of 1:100K (even 1:1M with a bit more sequencing), with high correlation with true tumor fractions 👉ultra-sensitive & quantitive monitoring, *without* deep targeted sequencing.

Similar methodology applied to copy number alterations expands this framework to more tumors with lower mutation rates.

MRDetect applied to pre-surgery CRC as ground truth showed high performance (with specificity to an individual patient). This allows to monitor treatment response to immunotherapy (melanoma), as well as detect post-operative residual disease, closely associated with outcome.

Applied to the challenging setting of lung adenocarcinoma (19% pre-operative detection with patient specific panels), we show maintained performance and post-operative residual disease detection.

Congrats to Asaf Zviran, @RafaelSchulman, @mjs2225 for leading this work! Huge thanks to many smart people who gave so generously to this labor of love. Special thanks to @wolchokj, @gmboland. To @notSoJunkDNA & @AltorkiNasser for believing in this work from day 1.

This work would not have been possible without the commitment of @TheMarkFdn, @LungAssociation, @WCM_MeyerCancer @nygenome for supporting high-risk work to impact patient care.

Finally, as always, we are standing on the shoulders of giants!

@CharlesSwanton @AshAlizadeh @velculescu Nitzan Rosendfeld @ldiaz1 @max_diehn @JShendure @MurtazaMDPhD and many more

access also at rdcu.be/b4ys4