Most cancers are detected by a cell biopsy, where a sample of the cancerous tissue is examined under a microscope. However, researchers in the US have reported that they are a step closer to developing a cancer screening tool that requires just a sample of blood.
The test, dubbed CancerSEEK, was developed by Johns Hopkins University, Baltimore. It looks for 16 DNA mutations commonly associated with cancer as well as eight proteins (biomarkers) associated with cancer.
CancerSEEK was tested on 1,000 patients with cancers in the ovary, liver, stomach, pancreas, oesophagus, colon, lung or breast. The cancers were at stage one to three and had not spread to other parts of the body (metastasised).
The scientists reported that the test was 70% accurate at spotting people with cancer. Importantly, the test raised few false alarms (“false positives”, in the terminology). In other words, it only “found” cancer in seven out of 812 healthy control subjects (people without cancer).
Important, but still hurdles to clear
This proof-of-concept study is important because it proves the feasibility of developing a fairly non-invasive and potentially inexpensive test in blood. And it is exciting because it confirms the usefulness of simultaneously looking for a combination of different molecules – such as DNA, RNA, proteins or metabolites – that are complementary and increase the likelihood of detecting cancer. This is a new approach and one that we have been developing for some time at Barts Cancer Institute for early detection of pancreatic cancer in urine.
The drawback of this study is that it is retrospective. Blood samples were taken from patients already diagnosed with cancer. And, although the test detects tumours that can be removed by surgery, they were not early stage tumours (only around 40% of stage one cancers were detected).
The test also appear to be least sensitive for two of the most common cancers (lung and breast), although this is probably due to the selection of biomarkers and can be further improved on. Another limitation is that the test doesn’t tell you where the cancer is located, but this could probably also be improved in future versions of the test by including of other variables in the model, such as symptoms or additional biomarkers.
As a proof of principle, this is an important and exciting study, but, before the new diagnostic tool is made available in hospitals and clinics, it will need to satisfy the requirements for any new test: rigorous further evaluation in large trials that would prove its effectiveness and usefulness as a cancer-screening tool.