How does CancerSEEK compare to other liquid biopsies?

This work[1] blows away previous efforts. The word “breakthrough” is used far too often to describe results that are incremental, or that are promising but not well-supported. CancerSEEK is a genuine breakthrough.

The gold standard for cancer diagnosis is the biopsy – a bit of suspect tissue is excised and studied, and determined to be benign or cancerous. The obvious limitation of this approach is that you have to detect the suspect tissue first. Detectable tumors are often advanced tumors, which are much harder to treat.

Finding blood-based markers of early-stage cancer has long been a goal of cancer diagnostics. But the best single blood marker for any cancer – PSA for prostate cancer – is not very good. High PSA levels are false-positive for cancer about 70% of the time, while low levels are false-negative about 15% of the time[2] . That’s terrible, and PSA testing is no longer recommended.

The liquid biopsy concept is based on the observation that tumors shed evidence of their existence into the bloodstream. This evidence can be the tumor cells themselves, their DNA and RNA, or their proteins. Tumor cells grow aberrantly and die at a high rate, and thus shed a fair amount of material into the blood.

Tumor cells can be used for liquid biopsies, but cells are hard to extract and analyze. DNA and proteins are fairly stable, and the technology to analyze them is very advanced. CellSEEK applies these technologies to detecting DNA alterations that are common in tumors and also to detecting aberrant proteins that are commonly produced by cancer cells. No single one of these markers has much diagnostic value alone, but combined, they are highly accurate and rich in information.

The key technological advance in this paper is to improve the accuracy of detection of cancerous DNA mutations while maintaining high sensitivity. This technique was combined with a systematic search through a set of cancer biomarker proteins to identify DNA-protein “signatures” that indicate cancer. And they did this work on a large and carefully characterized set of patients and healthy controls.

The result is that not only can CancerSEEK detect many types of cancer with high sensitivity, but can do so with very very few false positives. This last feature is incredibly important. Most people don’t have cancer, and so even a 2% false-positive rate will lead to more false-positives than true positives. The CancerSEEK false positive rate is 0.9%.

What’s even more impressive – and really qualifies this work as a breakthrough – is that the test works well on several cancers for which there are currently NO screening tests available. Ovarian, liver, and pancreatic cancers are typically not detected until very late stages, when they are essentially untreatable. We may well already havegood treatments for these cancers in their early stages, but we’ve never even tested them because we can’t identify patients early enough to give them a try.

This is possibly the biggest single advance ever in cancer diagnostics. The scope and the execution of the work are simply outstanding. I expect this test to get better and I expect it to become a routine part of healthcare in the next decade. This is what an actual breakthrough looks like.

Disclosure – I am a former employee and current shareholder in SomaLogic Inc, which develops blood protein signatures for clinical diagnostic and prognostic applications.


[1] Detection and localization of surgically resectable cancers with a multi-analyte blood test


Leave a Reply