The headline of this news release — ‘New blood test detects early stage pancreatic cancer’ — is misleading since the study results are quite preliminary, and the blood test was given to those already known to have cancer. For many readers the headline might suggest there’s a new screening test currently available that can detect this very lethal cancer in anyone, when that’s clearly not the case.
The strength of this news release is that it firmly establishes the need for early diagnosis in this difficult to diagnose condition.
Where it falls short is explaining what the main finding of detecting cancers with “96 accuracy” means, the significant limitations that place this impressive finding in a more appropriate context, and the potential influence of significant conflicts of interest among the authors.
A very small minority of cases of pancreatic cancer are caught early. By the time it’s usually diagnosed it’s frequently quite advanced and much more lethal. Any test — blood-based or otherwise — that could catch this cancer early has the potential to improve survival … and be lucrative for developers.
This news release teaches several lessons worth highlighting that we often find when blood tests are hyped as “catching cancer early.” First, by design, these tests are first applied (retrospectively) to people already known to have cancer. Just how useful they would be in screening the general public requires larger (prospective) studies. And this brings up a second lesson: The most important information for doctors and patients alike is how good is the test at detecting a cancer that’s there (specificity) and ruling out a cancer that’s not there (sensitivity). Thorough coverage of such blood tests needs to scrutinize the specificity/sensitivity, when possible.
[For further explanation on sensitivity and specificity see our toolkit article on the topic]
Finally, as is the case with this study, the financial stakes involved are significant. Much of the research in early cancer detection is industry-funded. Reporters covering such industry-backed tests should make this clear to the public, as well as seek independent sources to comment on the study design and results.
Cost is not mentioned.
Although the blood test is experimental, researchers would likely be able to provide a rough estimate of what such a 29-biomarker assay might cost.
From the lead author:
“Our test can detect pancreatic cancer with 96% accuracy at stage I and II while there is still the possibility of successful surgical intervention”
Using “accuracy” without context is misleading for two reasons. First, it’s not until the end of the news release that we learn that these impressive results are in people with already documented pancreatic cancer. Second, the 96% is somehow correlated with specificity (number of true cancers picked up by the test) and sensitivity (number of true cancers missed). That means 4% of those with cancer will be missed by the test (but the paper reports 6% false-positives; again, confusing).
Harms are not discussed and two types of harms are relevant here. The study — but not the news release — mentions a false-positive rate of 6%; that’s the fraction of people erroneously diagnosed with a cancer they don’t have. (Who might then go on to endure the costs and risks involved with subsequent evaluation.)
The flip side would be: what’s the false-negative rate? That’s the number of people who are told they’re cancer-free but really aren’t. The potential emotional harm in this case is quite clear.
This is important as well because pancreatic cancer is a rare condition. We know that a good screening test works best when the disease has high prevalence in a population, so when dealing with something rare, the 6% leading to harms may mean something very different.
It should have been made more clear to readers that the blood test was given to people already known to have pancreatic cancer. The findings are much too preliminary to suggest that such a test could be used as a clinical screening tool until it’s validated by larger scale prospective studies. This makes the news release headline potentially misleading.
This news release does not disease monger. It does quite well in establishing that an early detection method for pancreatic cancer is sorely needed since early clinical diagnosis is notoriously difficult with this very lethal cancer.
It should have been shared with readers that nearly half the study authors have financial ties to Immunovia AB, the biotech company developing this IMMray™ blood test for detecting pancreatic cancer.
The lead author quoted in the news release is a co-founder of the company and the chairman of its board of directors.
The release doesn’t mention how pancreatic cancer is currently diagnosed — CT scan, biopsy, carbohydrate antigen (CA) 19-9 blood test that measures the CA 19-9 level in the blood — and how this test might compare.
At the very end of the news release we read:
“In the future, the screening method could be used to screen people who are at high risk of developing pancreatic cancer, such as those with a hereditary risk, newly onset diabetes patients, and patients with chronic inflammation of the pancreas. The next step has already been initiated, which is a large US prospective study for high risk individuals.”
This is an important inclusion but it would have helped to put this higher up in the news release because it provides an opportunity to emphasize that the blood test is unproven as a screening tool.
More importantly, it’s currently not commercially available and the news release should have made this clear.
The release doesn’t specifically claim novelty, nor should it based on this very preliminary research.
There are currently two other blood tests that are marginally useful (when taken in the context of other tests) in helping to diagnose pancreatic cancer or monitor treatment: CA 19-9 and CEA.
There are no other existing blood tests that are routinely used by oncologists to diagnose pancreatic cancer.
The preliminary nature of these findings render the headline phrasing premature and unjustifiable.