[kazinator]

> The findings, based on samples from more than 2,500 tumours ...

Findings based on tumors sound like 20/20 hindsight!

If your crystal ball tells you where in a body tumor will develop five years from now, then if you look at the cells there, you might see some subtle signs that you know are not false positives.

Somehow, I don't think I'm holding my breath for new screening tests.

[sjg007]

The idea is to take blood samples from a patient population every year for a few years and sequencing the cells and cell free DNA population in the sample. You do this for years until the patient is diagnosed with cancer. At that point you can look back to the original samples a look for any potential biomarkers that in retrospect could have predicted the cancer. You do this for a large number of people. Right now we take a tumor and look for known biomarkers.

[skwb]

The other important thing to take into account is the vast heterogeneity of cancers. Cancers have been shown to be VERY diverse in their genomic expression profile, meaning you're probably looking at a very small pinhole compared to the entire tumor profile if you sample via biopsy. This diversity helps explains why cancers may initially respond to a specific targeted therapy, but then may reoccur with resistance initial therapy.