The rate of improvement for cancer liquid biopsies is astonishing. Two questions:

1. Based on your description of MRD and its lower tumor fraction, it seems technically more challenging than cancer screening?

2. In your view, what company(ies) shows the most promise in translating this type of research for commercial use in the clinic?

Expand full comment

Definitely, 300x improvements in signal enrichment in <2 years is very cool...

On a technical level, it seems like the tumor fraction is highest in patients with solid tumors BEFORE treatment. TF is lower and harder to detect after initial treatment. My understanding is that cancer screening in healthy individuals is a similarly challenging signal-to-noise problem.

For translation:

One contender is https://c2i-genomics.com/, which one of the PIs of the study is a scientific co-founder of.

I'm also curious about what will come out of Illumina's acquisition of Grail. As the sequencing market heats back up, I'm wondering if they are looking to more directly compete in screening, diagnostics, and other applications.

These are just some general thoughts, not an exhaustive analysis of the liquid biopsy market! Great questions.

Expand full comment

Yes definitely — you may be interested in giving this market analysis and predictions on GRAIL and the future of liquid biopsies a read: https://fundamentaldiagnosis.substack.com/p/illumina-part-4-durability-of-sequencing

C2i genomics is new to me. Freenome is another company I've been hearing.

Expand full comment

Great dive into the details.

I think doing a better job detecting residual disease would be a big advance if it lets less patients get treated (eg is more specific than current standard of care.) Doing really early detection of cancer before overt disease is also a big deal.

I wonder how the clinical trials for this stuff would look.

Expand full comment

Thanks Will!

Interesting point. Certainly, if MRD detection could be used to avoid unnecessary additional treatment, it seems like it would be a big clinical win.

I'm not as knowledgable about trial design for this, but one example can be found here for C2i Genomics (one of the PIs Dan Landau is a scientific co-founder)


Expand full comment

Amazing summary. It's great to see the study incorporate genomic context data (ATAC-seq, replication timing etc). The genomics/ML field is definitely headed multi-modal.

Would love to hear your thoughts on where ctDNA for early cancer detection is headed (particularly GRAIL).

Expand full comment