On 2014 Sep 03, David Keller commented:

Why physicians should intensify, not abandon, PSA screening

The recommendation to stop screening for prostate cancer is incorrect and will lead to a large increase in preventable cancer deaths. This monumental error was caused by the fallacy that the results of randomized, controlled trials are superior evidence to epidemiological observation, no matter how badly the trials were designed, no matter how contaminated the trial data, and without regard to the effects of intention-to-treat analysis on diminishing the measured benefits of screening. The 16 physicians and allied healthcare professionals on the USPSTF (United States Preventative Services Task Force) all hold positions of high esteem in primary-care specialties, and are expected to volunteer an average of about 4 hours of service per week. Incredibly, no urologists were enlisted to provide their crucial perspectives on PSA screening. The "D"-level recommendation against PSA screening was based on 2 poorly-designed and badly-executed clinical trials, which both generated highly corrupted data.

First of all, one must keep in mind the important epidemiological observation that, since the introduction of widespread PSA screening in the early 1990’s, there has been a 40% decrease in prostate cancer deaths. While even such powerful observational data cannot prove cause-and-effect, it also cannot be overturned by questionable results from flawed randomized trials. The two trials used by USPSTF do not withstand even casual scrutiny.

Both of the large randomized PSA screening trials were done in a manner which ignores the essential biological behavior of cancer cells, which is to grow and propagate faster than healthy cells. Both trials set a fixed threshold PSA level for biopsy (such as 4.0 ng/mL). Consider two men: the first man has a PSA of 3.9 in year 1 and a PSA of 4.1 in year 2. The second man has a PSA of 0.5 in year 1 and a PSA of 3.5 in year 2. Which man should have his prostate biopsied? Any rational physician would biopsy the second man, whose PSA rose by 700%, rather than the first man, whose PSA rose only about 5%. Yet, a study with a fixed biopsy threshold trigger of 4.0 would biopsy only the first man, who clearly is much less likely to have prostate cancer. By not including the rate of increase of PSA in the biopsy decision, both trials ignored information in the PSA which has the strongest correlation with the presence of prostate cancer.

In the U.S.-based PLCO trial, 52% of the men in the control group had off-protocol PSA screening by their private physicians, while only 86% of men in the screening group adhered to the full screening protocol. The men in the PLCO control group had an average of 2.7 PSA screening tests done off-protocol over the 6-year course of study (1). The purpose of PLCO was to compare outcomes in PSA-screened men versus non-screened men, which simply did not occur. The results of PLCO were distorted beyond recognition by mandatory intention-to-treat analysis, which assigned outcomes to each subject's original randomization group, regardless of what they did during the study. The majority of "control" subjects who actually received PSA screening by their private doctors were counted as unscreened, thus diminishing the apparent benefit of PSA screening.

The Europe-based ERSPC study, which had lower rates of protocol violations, produced data showing a significant decrease in prostate cancer mortality due to PSA screening. An ERSPC sub-study done in the city of Goteborg had a particularly low rate of off-protocol PSA testing, and showed correspondingly higher benefits in mortality reduction. The number needed to screen (NNS) to prevent one prostate cancer death at 14 years of follow-up was 293 in Göteborg, despite the fact that 24% of the men invited to be screened never showed up but were included in the intention-to-treat analysis, thus diminishing the measured benefit of screening (2).

Another flaw in the ERSPC study was the long interval between sequential PSA tests, which averaged about 4 years (3). Such a long interval of time between samples can allow the most aggressive, fast-growing cancers to evade detection while still confined to the prostate and curable. Men over 50 should have their PSA checked at least once a year for this reason, more often if fluctuations occur, which can be averaged to produce a better estimate of the baseline PSA, deviations from the baseline, and the rate of increase of PSA. The noisier a signal, the more data samples are required to characterize it accurately.

Prostate cancer kills more men every year than any other cancer except lung cancer. It is a serious threat to all men over age 50 with a life expectancy of greater than 10 years. We know that it is curable when confined to the prostate. More PSA testing, not less, is called for. Improved diagnostic techniques, such as high-resolution prostate MRI imaging, and numerical filtering of PSA values, are being developed. One final randomized trial, which includes all of these improvements, must be conducted to obtain a fair assessment of PSA screening. The problem will be to find enough men willing to forgo screening to fill the control group.

In general, a man whose PSA is 0.1 has little to fear from prostate cancer, and a man whose PSA is 100 has much to fear from prostate cancer. The problem is to clearly and accurately determine how to proceed within those 3 orders of magnitude of data. The art of medicine is to act in the best interests of the patient when the scientific facts are not perfectly understood. Clinical common sense and scientific intuition demand that we measure and carefully consider the PSA levels of men over the age of 50, pending the results of further trials. Watchful waiting is always an option, which at least provides prognostic awareness.

The results of poorly-designed and poorly-executed controlled clinical trials give the mere illusion of high-quality data. Their results should not be relied upon as a basis for treatment decisions when they contradict powerful epidemiological data and our understanding of the biological basis of the disease.

References

1: Gulati R, Tsodikov A, Wever EM, Mariotto AB, Heijnsdijk EA, Katcher J, de Koning HJ, Etzioni R. The impact of PLCO control arm contamination on perceived PSA screening efficacy. Cancer Causes Control. 2012 Jun;23(6):827-35. doi: 10.1007/s10552-012-9951-8. Epub 2012 Apr 10. PubMed PMID: 22488488; PubMed Central PMCID: PMC3556907.

2: Hugosson J, Carlsson S, Aus G, Bergdahl S, Khatami A, Lodding P, Pihl CG, Stranne J, Holmberg E, Lilja H. Mortality results from the Göteborg randomised population-based prostate-cancer screening trial. Lancet Oncol. 2010 Aug;11(8):725-32. doi: 10.1016/S1470-2045(10)70146-7. Epub 2010 Jul 2. PubMed PMID: 20598634; PubMed Central PMCID: PMC4089887.

3: Schröder FH, Hugosson J, Roobol MJ, Tammela TL, Ciatto S, Nelen V, Kwiatkowski M, Lujan M, Lilja H, Zappa M, Denis LJ, Recker F, Berenguer A, Määttänen L, Bangma CH, Aus G, Villers A, Rebillard X, van der Kwast T, Blijenberg BG, Moss SM, de Koning HJ, Auvinen A; ERSPC Investigators. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med. 2009 Mar 26;360(13):1320-8. doi: 10.1056/NEJMoa0810084. Epub 2009 Mar 18. PubMed PMID: 19297566.

On 2014 Sep 03, David Keller commented:

Why physicians should intensify, not abandon, PSA screening

The recommendation to stop screening for prostate cancer is incorrect and will lead to a large increase in preventable cancer deaths. This monumental error was caused by the fallacy that the results of randomized, controlled trials are superior evidence to epidemiological observation, no matter how badly the trials were designed, no matter how contaminated the trial data, and without regard to the effects of intention-to-treat analysis on diminishing the measured benefits of screening. The 16 physicians and allied healthcare professionals on the USPSTF (United States Preventative Services Task Force) all hold positions of high esteem in primary-care specialties, and are expected to volunteer an average of about 4 hours of service per week. Incredibly, no urologists were enlisted to provide their crucial perspectives on PSA screening. The "D"-level recommendation against PSA screening was based on 2 poorly-designed and badly-executed clinical trials, which both generated highly corrupted data.

First of all, one must keep in mind the important epidemiological observation that, since the introduction of widespread PSA screening in the early 1990’s, there has been a 40% decrease in prostate cancer deaths. While even such powerful observational data cannot prove cause-and-effect, it also cannot be overturned by questionable results from flawed randomized trials. The two trials used by USPSTF do not withstand even casual scrutiny.

Both of the large randomized PSA screening trials were done in a manner which ignores the essential biological behavior of cancer cells, which is to grow and propagate faster than healthy cells. Both trials set a fixed threshold PSA level for biopsy (such as 4.0 ng/mL). Consider two men: the first man has a PSA of 3.9 in year 1 and a PSA of 4.1 in year 2. The second man has a PSA of 0.5 in year 1 and a PSA of 3.5 in year 2. Which man should have his prostate biopsied? Any rational physician would biopsy the second man, whose PSA rose by 700%, rather than the first man, whose PSA rose only about 5%. Yet, a study with a fixed biopsy threshold trigger of 4.0 would biopsy only the first man, who clearly is much less likely to have prostate cancer. By not including the rate of increase of PSA in the biopsy decision, both trials ignored information in the PSA which has the strongest correlation with the presence of prostate cancer.

In the U.S.-based PLCO trial, 52% of the men in the control group had off-protocol PSA screening by their private physicians, while only 86% of men in the screening group adhered to the full screening protocol. The men in the PLCO control group had an average of 2.7 PSA screening tests done off-protocol over the 6-year course of study (1). The purpose of PLCO was to compare outcomes in PSA-screened men versus non-screened men, which simply did not occur. The results of PLCO were distorted beyond recognition by mandatory intention-to-treat analysis, which assigned outcomes to each subject's original randomization group, regardless of what they did during the study. The majority of "control" subjects who actually received PSA screening by their private doctors were counted as unscreened, thus diminishing the apparent benefit of PSA screening.

The Europe-based ERSPC study, which had lower rates of protocol violations, produced data showing a significant decrease in prostate cancer mortality due to PSA screening. An ERSPC sub-study done in the city of Goteborg had a particularly low rate of off-protocol PSA testing, and showed correspondingly higher benefits in mortality reduction. The number needed to screen (NNS) to prevent one prostate cancer death at 14 years of follow-up was 293 in Göteborg, despite the fact that 24% of the men invited to be screened never showed up but were included in the intention-to-treat analysis, thus diminishing the measured benefit of screening (2).

Another flaw in the ERSPC study was the long interval between sequential PSA tests, which averaged about 4 years (3). Such a long interval of time between samples can allow the most aggressive, fast-growing cancers to evade detection while still confined to the prostate and curable. Men over 50 should have their PSA checked at least once a year for this reason, more often if fluctuations occur, which can be averaged to produce a better estimate of the baseline PSA, deviations from the baseline, and the rate of increase of PSA. The noisier a signal, the more data samples are required to characterize it accurately.

Prostate cancer kills more men every year than any other cancer except lung cancer. It is a serious threat to all men over age 50 with a life expectancy of greater than 10 years. We know that it is curable when confined to the prostate. More PSA testing, not less, is called for. Improved diagnostic techniques, such as high-resolution prostate MRI imaging, and numerical filtering of PSA values, are being developed. One final randomized trial, which includes all of these improvements, must be conducted to obtain a fair assessment of PSA screening. The problem will be to find enough men willing to forgo screening to fill the control group.

In general, a man whose PSA is 0.1 has little to fear from prostate cancer, and a man whose PSA is 100 has much to fear from prostate cancer. The problem is to clearly and accurately determine how to proceed within those 3 orders of magnitude of data. The art of medicine is to act in the best interests of the patient when the scientific facts are not perfectly understood. Clinical common sense and scientific intuition demand that we measure and carefully consider the PSA levels of men over the age of 50, pending the results of further trials. Watchful waiting is always an option, which at least provides prognostic awareness.

The results of poorly-designed and poorly-executed controlled clinical trials give the mere illusion of high-quality data. Their results should not be relied upon as a basis for treatment decisions when they contradict powerful epidemiological data and our understanding of the biological basis of the disease.

References

1: Gulati R, Tsodikov A, Wever EM, Mariotto AB, Heijnsdijk EA, Katcher J, de Koning HJ, Etzioni R. The impact of PLCO control arm contamination on perceived PSA screening efficacy. Cancer Causes Control. 2012 Jun;23(6):827-35. doi: 10.1007/s10552-012-9951-8. Epub 2012 Apr 10. PubMed PMID: 22488488; PubMed Central PMCID: PMC3556907.

2: Hugosson J, Carlsson S, Aus G, Bergdahl S, Khatami A, Lodding P, Pihl CG, Stranne J, Holmberg E, Lilja H. Mortality results from the Göteborg randomised population-based prostate-cancer screening trial. Lancet Oncol. 2010 Aug;11(8):725-32. doi: 10.1016/S1470-2045(10)70146-7. Epub 2010 Jul 2. PubMed PMID: 20598634; PubMed Central PMCID: PMC4089887.

3: Schröder FH, Hugosson J, Roobol MJ, Tammela TL, Ciatto S, Nelen V, Kwiatkowski M, Lujan M, Lilja H, Zappa M, Denis LJ, Recker F, Berenguer A, Määttänen L, Bangma CH, Aus G, Villers A, Rebillard X, van der Kwast T, Blijenberg BG, Moss SM, de Koning HJ, Auvinen A; ERSPC Investigators. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med. 2009 Mar 26;360(13):1320-8. doi: 10.1056/NEJMoa0810084. Epub 2009 Mar 18. PubMed PMID: 19297566.