Well … Don’t say we didn’t warn you. The American Cancer Society (ACS) and others are now saying that the study suggesting a rise in the incidence of metastatic prostate cancer from 2004 to 2013 has serious flaws.
You can read this article in the New York Times for yourselves. Basically, the ACS is indicating that there are significant statistical and epidemiological problems with the data presented in the paper by Weiner et al. that we commented on yesterday, and the ACS seems to agree with us that at least some of the rise in incidence could well be associated with improvements in the ability to detect metastasis earlier because of better imaging.
We would simply repeat our caution from yesterday that one has to be extremely careful in how one analyzes and interprets data of this type.
Filed under: Diagnosis, Risk | Tagged: Diagnosis, incidence, metastasis, risk |
THE "NEW" PROSTATE CANCER INFOLINK 
I do not see a reference to improved screening in the New York Times article, although it may have been stated on the ACS website; I agree, it is a valid point.
The New York Times article does quote Otis Brawley contesting the study’s epidemiological validity — although it never seems to have troubled Dr. Brawley before, since he oft quotes absolute numbers.
What the New York Times reporter does not report is that Dr. Brawley has a strong bias against PSA testing that has carried through much of ACS’ public support for men living with prostate cancer by reducing services. Dr. Brawley publicly boasts he does not know his own PSA as an African-American over the age of 50.
While the NBC reporter considers that “PCa is a huge issue in this country”, one would never know it from the ACS. By way of example … has anyone ever seen ACS promoting prostate cancer awareness at a Relay4Life event? I recently attended an ACS-sponsored conference that had more than 30 posters addressing breast cancer and less than 6 focused on prostate cancer.
Anyone who believes anything Dr. Brawley says about prostate cancer will likely believe Donald Trump’s acceptance speech earlier tonight.
I decided to look into the Sitemaster’s hypothesis that improved technology and its increased use contributed to the increase in metastatic diagnoses. I reasoned that, if true, many cases that would formerly have been diagnosed as high risk would now be diagnosed as metastatic. So I simply added together the number of high-risk and metastatic cases over the years.
High-risk diagnoses rose rapidly until 2008, and then fell sharply to a nadir in 2012 and about the same in 2013. Metastatic diagnoses reached a nadir in 2007, and then increased steadily through 2013. Adding them together, the overall pattern is flat. There were a few years (2007-8 and 2010-11) where the combined total was over 18,000 cases per year, but the other years averaged around 17,000 cases. This seems to support the Sitemaster’s hypothesis.
While the total number of high-risk plus metastatic cases is pretty steady, their percent of all diagnosed cases has risen from 22% in 2011 (range was 21% to 23% in 2004 to 2011) to 24% in 2012 to 26% in 2013. This is exactly the pattern we hope to see if we have a more rational screening policy — the number of diagnosed higher-risk cases stays flat while the number of diagnosed favorable-risk cases declines.
The data in their analysis seem to support the hypothesis of improved screening technology and policy over the last few years.
Thanks for doing the heavy lifting Allen.
My observation is that the reduction in overall diagnoses is attributable to reduced PSA testing … so the increased proportion for high-risk and metastasized disease is deceptive. I do agree that improved screening has allowed the maintenance of better diagnosis for high-risk and Mx disease.
I am, however, somewhat confused by your percentages. For example, in 2013 the CDC reports a total of 176,540 diagnosed men … +/- 17,000 represents 10%. What am I missing?
Also, where do I find diagnoses by risk category?
Rick:
The increased proportion is not deceptive. If the reduction in overall diagnoses were consistent, we would not see an increase in the high-risk/metastatic group relative to the favorable-risk patients — all risk groups would have been affected similarly. Moreover, the diagnoses of high-risk/metastatic patients remained fairly constant. At the same time, low-risk diagnoses fell from 37% of all new cases in 2004 to only 24% of all new cases in 2013 — a decrease of 37% across the 10-year period. Therefore, we are consistently finding the most unfavorable-risk patients while missing only the low-risk patients. This is exactly the pattern we want to see: a reduction in over-diagnoses, and a consistent diagnosis of the most unfavorable-risk cases.
What you are missing is that you are looking at different databases. The NCDB is what was used in this study and accounts for only those men who were newly diagnosed with prostate cancer at the 1,089 Commission On Cancer accredited facilities. The CDC number is a projection for the full US based on their algorithm. The incidence by risk group was reported in Table 1 in the study discussed by the Sitemaster.
Thanks for clarifying the difference in the databases Allen.
I still believe all testing is now innately biased, and therefore deceptive. … While less men are tested, more are high-risk/advanced because they may be symptomatic in some mild form and are seeking testing as a result of more advanced disease.
In other words, PSA testing is now biased to testing men who are either at higher risk or symptomatic/advanced.
Isn’t that exactly the “bias” we want to see? We want to identify the men who need treatment, and leave those who need no treatment unmaimed.
Agreed … but my point is that it distorts and overstates the proportion of high-risk and metastatic men compared to the whole population who have prostate cancer.