Blood Tests For Cancer

"Liquid biopsies" for cancer are coming to the clinic.

Cancer remains the winner in the war on cancer. New molecularly-driven precision treatments have improved outcomes for a few types of cancer, and the reduction in smoking has provided substantial improvements in death rates, but the overall statistics remain grim, most treatments are dreadful, and early detection is more a mirage than reality. One promising, though still experimental, area of progress is in detecting cancers using blood samples.

Cancer trends in the US, overall.

Early detection has been a holy grail, with enormous resources devoted to mammography and PSA tests, among much else, which have turned out to be of marginal utility, or far less than touted. I do not believe there is currently any cancer for which a reliable medical test of any kind provides detection before symptoms or manual / visible detection is possible. After the various reliable and unreliable methods of detection, assessment of cancers involves biopsy, which is far more invasive and disruptive than it sounds, piercing the putative site / organ with a large sampling needle which can cause permanent damage. Biopsy should be regarded as a full surgical procedure in its own right.

Both of these problems could be alleviated with effective blood tests for cancer presence, type, and progression. A significant development in the research field over the last decade or two has been the realization that cancers shed material constantly. Cells are sloughed off in live and dead form, and DNA from tumors is generally in circulation. One corollary is that metastasis is more a matter of these cells finding a congenial home than of their dispersal from their primary source. A second is that blood tests can detect these DNAs and cells on a routine basis.

The root method for doing so is PCR- that revolutionary method in molecular biology that harnesses DNA replication to amplify nucleic acids exponentially, allowing detection of infinitesimal amounts. One of the papers under review in this post claims that a single molecule of cancer cell DNA can be detected in 5 ml of blood. This is astonishing, but also puts bounds on the ultimate utility of this method, since they also say that less than half of grade 1 cancers provide even such a tiny signal. It turns out that, as one might expect, earlier and smaller cancers shed less material than later ones do.

Early stage cancers are hard to detect, but not impossible. The lowest Y-axis levels correspond to one molecule in the sample.

This landmark paper tests patients with many different types of cancer to evaluate the possibility of a relatively blood test for certain known cancer mutations. They find that brain cancers are particularly poorly represented- their shed materials are likely to be confined due to the blood-brain barrier system, plus the glymphatic system. But other cancers are quite amenable to blood testing, at least when in an advanced state. This would at least be a boon to recurrence tracking, and treatment monitoring, for which (repeated) biopsy is either impractical or impossible.

Which cancers give usable blood-born DNA samples?

"... 47% of patients with stage I cancers of any type had detectable ctDNA, whereas the fraction of patients with detectable ctDNA was 55, 69, and 82% for patients with stage II, III, and IV cancers, respectively."

For early screening, blood testing is not, as of this paper in 2014, truly reliable. On the other hand, it finds half of stage 1 cancers, which otherwise might not be found at all, raising the question of how such a cancer should diagnosed and found if a blood test finds, for example, that a common mutation (for example, in the gene TP53) is found to be afoot in a patient. Such mutations, which drive many different cancers, could come from virtually any organ. Some more sleuthing would be in order.

One such approach came up recently, in studies of regulatory markings on DNA, which some call "epigentic" marks. C nucleosides in DNA can be methylated and then derivitized from there to 5-hydroxymethyl 5-formyl, 5-carboxyl, and finally identified by the DNA repair pathway and excised / replaced. Typically, methylation is a repressive signal, part of the cellular machinery that turns off gene expression. In contrast, 5-hydroxymethy modified C residues seems to be associated with higher gene expression. At any rate, both modifications are dramatically reduced in cancer cells, and their patterns can be informative about the cancer's tissue of origin and prognosis/stage. There is even the possibility that the relative positions of 5-methyl-C and 5-hydroxymethyl-C in very small segments of DNA (detected by FRET, no less) could be informative on these issues, though that is more esoteric.

So far, these methods are plumbing the blood samples for specific DNA mutations in specific genes known to drive cancer, and thus have high specificity, but limited utility as general screening tools for patients who have not yet been diagnosed and could have any (or several) of thousands of different mutations. To do that, a far larger panel of genes needs to be assayed, possibly even whole genome sequencing, with an unbiased analysis of their mutations. But that begs the question of how to separate the cancer-derived DNA from all the other junk floating around in a blood sample. Methylation marks may be biased in cancer-derived DNA in useful ways, but they do not have categorically different characteristics usable for separating the wheat from the chaff. This is the big problem right now in cancer blood testing. On a practical level, it will start being used for already-diagnosed patients, to track their treatment and relapse. The cancer selection problem will likely be solved in a brute-force way by sequencing everything in the blood sample and sifting through that data using a growing catalog of cancer-causing mutations. But if some mark or characteristic can be found that is specific to cancer DNA, then general and convenient cancer screening via blood tests will come much sooner.

• Rules are for little people.
• Such as immigrants. Who are Christians, right?
• Does Amazon need to be split up?
• Go ahead and stretch ... every day.
• Skateboard therapy.
• More bullying and sleaze.
• Studies in groveling and sycophancy. What causes this temperament?
• Remember Martha Mitchell?
• The Taliban is winning.
• Meanwhile, Japan has a problem of not asking questions.
• Capitalism is so not working.
• Aretha on piano.
• In California, old-money landlords make out like bandits.