A news release from the University of York describes a new test that uses computer algorithms and genetic markers to distinguish between aggressive and more manageable types of prostate cancer. While the study is informative, the news release summarizes it in confusing terms and analogies that may be difficult for a lay reader to understand. Furthermore, it neglects to mention until the very end of the release that this test has only been done in the laboratory, and is likely still years away from being available to actual patients.
Prostate cancer is relatively common. According to the American Cancer Society, 1 in 9 men will be diagnosed with prostate cancer in their lives. It is also the second leading cause of cancer deaths in American men. Some prostate cancers are more aggressive than others, however. A test that could distinguish between aggressive and more manageable cancer types could help some patients avoid unnecessary and risky radiotherapy and surgery.
An estimated 40% of prostate cancers detected by the standard PSA (prostate-specific antigen) test are considered over-diagnosed. For decades, the great majority of these men were aggressively over-treated with surgery or radiation. In recent years, more men with low-risk cancers are opting for active surveillance, a monitoring strategy that involves repeated PSA tests and biopsies. Only men with evidence of cancer progression will be offered curative treatment. However, the current methods for determining low-risk status — PSA and Gleason score — are considered relatively crude. A more accurate test would not only help limit over treatment but would potentially give men more confidence in selecting active surveillance.
This news release does not mention costs. This is likely because, as we said above, the test is still experimental, and not something available to the general public.
In comparison, available genomic tests run around $4,000.
The news release tries to quantify benefits in two ways. First, it says that “A recent study showed that more than 25 men were being unnecessarily treated with surgery or radiotherapy, for every single life saved.” But how recent was this study? And who published it? Next, the release says that the news test “can pick out life-threatening prostate cancers, with up to 92% accuracy,” although this is not technically correct. In the published study, the authors say that the algorithm resulted in “an AUC of 0.92.” AUC stands for Area Under the Curve, which is not the same measurement as accuracy.
The AUC is considered the best measure of whether a test can distinguish between diseased and non-diseased people. In this case, the investigators used their test to differentiate between cancers that were confined to the prostate and those that had spread beyond the prostate among men who had undergone prostatectomy. While these advanced cancers can be life-threatening, that’s a bit of hyperbolic description. Metastatic cancers are clearly the most likely to be life threatening. Additionally, estimates of AUC have 95% confidence intervals. If the AUC, as it appears, was based on results from 4 patients, then 92% is a very imprecise estimate.
The news release says that unnecessary radiotherapy and surgery is harmful, but it does not explain any harms of the new tests. Although biopsies are minimally invasive, the Mayo Clinic says that prostate biopsies can cause bleeding and risk of infection.
In this test, the tissue was collected only from prostatectomy specimens. Therefore, to actually be useful for preventing over treatment, the test must be assessed in biopsy specimens. The harm would be misclassification — mistakenly telling someone that treatment is unnecessary because the tumor appears to be low-risk when in fact it’s aggressive and would benefit from treatment.
Neglecting to provide the number of patients and failing to explain the meaning of 92% accuracy, made this an evidence-free release. (Information on some of the challenges of describing accuracy, sensitivity and specificity of diagnostic tests is available in this HealthNewsReview.org toolkit.)
After reading the study, it seems that biopsies were only taken from four patients, a very small sample size. The news release also attempted to summarize the methods of the study, but ended up doing so using confusing language and analogies that were hard to parse.
The term “life-threatening cancer” was used very loosely. A cancer that has spread beyond the prostate — but not throughout the body–is not necessarily fatal.
All funding sources were listed at the bottom of the release.
The alternatives to this test include other genomic tests; PSA, Gleason scores from biopsies, and MRI results, none of which are mentioned.
The release does say that this test could be an alternative to radiotherapy and surgery (by distinguishing non-aggressive from aggressive cancers), but it doesn’t list any of the harms or benefits of other interventions. It also does not say what treatment options there are should a patient find that they have a non-aggressive form of cancer.
At the very end of the news release, the authors write that the next step is, “to take this method out of the laboratory,” and that the researchers hope that they will find a commercial partner to develop the test for general use. But this should have come much sooner, ideally in the first paragraph. The entire release (including the headline) made it sound like this test is already available, likely frustrating patients who felt duped at the end of the release.
The news release does attempt to describe the novelty of the test, although the release struggles to describe scientific terms.
No sensational or unjustifiable language.