The headline scrolling across the bottom of an evening news show certainly grabbed my attention: a new blood test had the possibility of detecting early melanoma and saving thousands of lives. And then there were more reports elevating this early research report to a point that I became quite interested—and frankly concerned.
News flash: The research is far from being shown to have proven value in the early diagnosis of melanoma, a less common but certainly potentially deadly form of skin cancer. I am not certain the various media headlines and articles appropriately reflected that fact, once again offering hype over inconvenient reality.
The researchers who wrote the report are from Australia, a country with a particularly high incidence of melanoma and a long history of trying to increase public awareness to prevent melanoma and alternatively find it early, when it is most effectively treated.
Basically, and in very simple terms, the researchers compared blood samples from a group of 104 early stage melanoma patients and 105 healthy volunteers, then validated their results among a smaller group of 20 early stage patients and 16 healthy volunteers. They then analyzed these blood samples against a panel of 1627 protein markers in the blood, and eventually settled on a panel of 10 of these markers which—when taken together as a group—appeared to signal the presence of a melanoma when comparing the samples from the patients known to have melanoma to the patients who did not have the cancer.
The “scores” for the melanoma patients were higher than for the healthy controls, as noted by the authors. These 10 markers in combination were correctly “positive” (called “sensitivity”, that is finding a disease when it is present) in close to 8 out of 10 melanoma patients. That means it would miss 2 out of 10 melanomas). The test also correctly identified 85% of the healthy participants as not having melanoma (called “specificity”), which –importantly—means 15% of the patients who did not have melanoma were tagged as having the disease present when it wasn’t there).
That’s actually a pretty good showing for a new test that has not been rigorously evaluated. And that’s where the somewhat misleading news reports come in. This is not a test ready for general use. There is a lot more research to be done on much larger groups of patients to see if it really works as an “early detection” test. I guess it could work, but it’s a long way from what the headline implied when it blared “Experimental blood test could detect melanoma skin cancer early, study finds.” My friends, I assure you that journey is one that will take probably years to travel to find out whether this test is the real deal and most important makes a real difference in the outcomes for the people who do develop melanoma.
To be honest, what I find so interesting is that this test was “positive” in patients who had very early melanomas, some I assume that were so small they would be hard to identify as melanoma. The actual surface measurements of the cancers weren’t provided (which would give some idea of how large they were), although there was some data regarding basic features of the melanomas studied: three quarters of the patients had melanomas that were less than 1 mm deep into the skin; 43% were “in situ” melanomas, which were not invasive and theoretically not at risk of spreading.
You may not be a melanoma expert, however if you happen to know one they may well tell you those are early melanomas. And yet these early melanomas were apparently able to generate a complex antibody response that could be detected with a fairly high degree of accuracy according to the research.
Let’s just say that’s something I need to learn more about, namely how a very small lesion can produce enough antibody protein in the blood such that it would distinguish one person with a very small, early melanoma from another who did not have melanoma. I find that very intriguing. I am not aware of any commonly used blood-based protein test that is comparable that can make a claim to detect such a small amount of cancer with such a high degree accuracy to discriminate it from people who don’t have cancer. We are talking about millimeters here, not inches or ounces of tumor. I am not saying it can’t happen, just that I find it fascinating that it does.
It’s also important to note these blood tests were done in patients whose melanomas had already been diagnosed through routine diagnosis and excision, within one month after that diagnosis and in some cases as long as three months later—after the tumor was removed. And no patients with other cancers were tested to see how specific these results were to melanoma vs. possibly other cancers as well. As noted below, the authors also make that point: at this juncture, it’s not certain these test results are necessarily specific to melanoma or even other illnesses.
Let me be clear here: I am not critical of the scientists’ methods or reports. I accept them at face value. However, I am critical of some in the media promoting this research as something that will in short order revolutionize the early detection of melanoma. That may happen one day, but not today and not tomorrow. And there is certainly no guarantee that it will happen, as history has shown us time and again.
I would like to close with some comments taken directly from the authors’ manuscript because I think they say exactly what needed to be said and should be shared with the public:
“Most importantly, the panel identified here must be validated using larger cohorts of melanoma patients and should include patients with other types of cancer or autoimmune diseases to ascertain whether the combination is melanoma specific.”
The problem, of course, is that level of caution doesn’t make for catchy headlines and doesn’t engender false optimism.
Time and again, we have seen the promise of early detection in some cancers crash on the shores of evidence. It’s time we learned that these “breakthroughs” require painstaking research both in the lab and the clinic before we really know whether they make a difference. And this test is no exception.