When it comes to cancer, I used to be convinced that early detection was where most of our efforts should go. I was appalled that so much money was being poured into developing therapeutics for patients with end-stage cancer. Surely the trick was in detecting the cancer early enough to prevent all of this.
A new study came out in the journal Science about a blood test that can detect eight different cancers. It’s called CancerSEEK and its pedigree is certainly impressive, in terms of the publishing journal, the academic centres involved, and the researchers who authored this article. Fun fact about one of its senior authors, Bert Vogelstein: he is the 9th most cited researcher according to Google Scholar citations, well ahead of Albert Einstein (at number 1,491), but trumped by Sigmund Freud at number 1.
CancerSEEK is the latest iteration in a quest for oncological gold. Wouldn’t it be great if we could draw your blood and tell you, with absolute certainty, not only if you have cancer, but which cancer? Some media outlets certainly seem to think CancerSEEK is the Holy Grail. The Baltimore Sun called it a “cancer breakthrough”. The Guardian was less overt, but note the loaded language (emphasis in italics mine): “New blood test could help detect eight common cancers before they spread: Researchers believe CancerSEEK will save thousands of lives and hope it will be widely available in a few years.”
For CancerSEEK to be valuable in a clinical setting, it needs high sensitivity and specificity. A cancer test that is 100% sensitive will correctly identify every person who has cancer as having cancer. If the test has too low a sensitivity, it will miss people who have cancer and who might benefit from treatment.
Meanwhile, a cancer test that is 100% specific will correctly identify every person who does not have cancer as not having cancer. If the test is not very specific, it will flag as sick people who aren’t, and this leads to more invasive diagnostic procedures and, sometimes, to unnecessary treatments.
How does CancerSEEK fare? When the test was used on 812 healthy controls, only 7 were flagged by the test as having cancer, which leads to a reported specificity above 99%. We need to keep in mind that this was done in a fairly small sample of individuals. This preliminary result is great but what about its sensitivity, or its ability to actually detect cancer when it is there? This is where the results, shall we say, vary tremendously. It did very well with cancers of the ovary and the liver, not so well with stomach, pancreas, esophagus, colorectum, and lung… and could only detect 33% of breast cancers. Moreover, when it came to detecting early cancers, CancerSEEK did no better than flipping a coin.
Beyond issues of sensitivity and specificity, early detection has another problem that was not clear to me when, years ago, I embraced the concept unconditionally.
Just because you have cancer cells doesn’t mean cancer will kill you.
We’ve been through this before with PSA testing for prostate cancer. An elevated level of PSA in the blood was flagged as abnormal, and many men received surgery and/or radiation to treat something that never would have become a concern otherwise. These treatments often come at a price, as Dr. Gilbert Welch writes: “Roughly a third suffered side-effects of treatment -- generally related to bowel, bladder or sexual function. And a few have died from it.”
Some cancers grow rapidly and must be taken out; others grow very slowly and at such a late age that watchful waiting is often wiser. And that’s the problem with early detection. It sounds like more of it is always good, but sometimes the diagnosis itself leads to more harm: psychological stress, physical side effects, and financial burden. The Choosing Wisely campaign, which fosters a dialogue between clinicians and patients on the issue of unnecessary medical tests, is a step in the right direction.
Dr. Welch made a good point in his op-ed about CancerSEEK: “Americans don’t need more diagnoses, they need the right diagnoses.” Whether the test can be refined to improve its sensitivity remains to be seen. Many such tests failed to move beyond their early, promising results and ended up in the dustbin of biomedical research. Even if CancerSEEK can be improved, its indiscriminate use as a screening tool for the masses risks creating more problems than it will solve. When it comes to medical tests, we should all remember that, often and somewhat counterintuitively, less is better.
Want to engage with this content? Comment on this article on our Facebook Page!