Revisiting my personal decision about prostate-specific antigen testing in 2005



Somewhat atypical for older male American primary-care physicians, amongst whom almost 80% have made the personal decision to have a PSA test [1], I have not. However, there is new evidence to consider as I ponder whether I should join the ranks of so many of my colleagues and be screened for prostate cancer, as I approach the age of 52 years in a few weeks.

First, some background: I am Caucasian and have no first-degree relatives with prostate cancer; I have minor LUTS (IPSS of 6); I have frequent bone pain, but those symptoms are migratory and evanescent, and seem easily attributable to attempting to maintain my exercise level despite the ageing process. As such, my lifetime probability of dying from prostate cancer is ≈ 3% (even using statistics from before the advent of PSA testing), although my lifetime probability of dying is, sadly and inevitably, 100% (with the ultimately tight CI). Moreover, if I am destined to die from prostate cancer, there is a 70% chance I will do so after age 75, more than two decades hence.

What new evidence should I be considering as I revisit my personal decision about PSA testing? First and foremost are the updated results from the SPCG-4 trial on radical prostatectomy vs watchful waiting [2]. This important trial provides the first evidence that for men with clinically localized prostate cancers that are largely discovered the ‘old fashioned’ way, without PSA screening, surgery can change the natural history of the disease. However, the absolute benefits appear to be fairly small, so that ≈ 20 men would need to have their prostates removed to prevent one death over 10 years. How would this ‘number needed to treat’ (NNT) change for men with cancer discovered by screening? The effects of over-diagnosis and lead time would probably mean that ≈ 40 men would need to have surgery to save a life over 15–20 years, which sounds like a very small benefit to me. However, the NNT would shrink if earlier detection saved some lives of men who would otherwise have died in the surgical arm of the SPCG-4 trial, or even some men not included in the trial because they initially presented with high-grade or disseminated disease. Unfortunately, we do not have the data to quantify that effect on the NNT, as yet. Moreover, any benefit would come at the price of a substantial risk of sexual dysfunction and incontinence, as documented in the trial; neither prospect sounds attractive to me.

The work of D’Amico et al.[3] is also a key new piece to the puzzle. These investigators documented that there is a group of patients with prostate cancers characterized by rapidly rising pre-diagnostic PSA levels that, even when found relatively early and treated aggressively, do not fare well, with a 15–20% risk of dying of prostate cancer within 10 years. Although we cannot determine whether such men would have fared even worse without early detection and treatment, these data suggest there is a subset of cancers that may simply not be amenable to screening, and may well be responsible for a disproportionate share of prostate cancer mortality.

Another important piece of data for me is the newest estimate of the lifetime risk of a prostate cancer diagnosis in the USA, where PSA screening is widespread. In the ‘pre-PSA era’, the estimated lifetime risk of a prostate cancer diagnosis for a man my age in the USA was ≈ 10%. In the ‘PSA era’, as of 2000–2002, it was 19%[4], and not all men have been tested; it would easily be >20% for a man who has a regular PSA test. That risk will be driven relentlessly higher in future years as the PSA threshold for biopsy is lowered and increasingly many biopsy cores are taken. I suspect that if epidemiological studies indicated that regular consumption of a certain vegetable doubled the risk that a man would ‘get’ prostate cancer, no one would eat it. The PSA test is that vegetable. Food for thought.

The lifetime probability of requiring one or more sets of prostate biopsies will be much higher than the lifetime probability of a cancer diagnosis, and that prospect does not thrill me either. That cumulative risk has not been quantified, but the poor discriminating ability of PSA suggests that it will be high. Data from the Prostate Cancer Prevention Trial have allowed the first unbiased receiver-operating characteristic curve (ROC) to be drawn for the PSA test, using biopsy as the diagnostic ‘gold standard’. The ROC area for PSA (the chance that a man with prostate cancer on biopsy will have a higher PSA level than a man without) is only a mediocre 68%[5], and not much better than that of flipping a coin (50%). If we feel we must biopsy at relatively low PSA levels to maintain adequate sensitivity, the number of biopsies that will be required given this poor discrimination will be numerous indeed.

As I contemplate these new data, I am left fairly certain that if I elect to start PSA testing now, I will at least double my risk of having to deal with prostate cancer over my lifetime, from ≈ 10% to ≈ 20%, and probably even higher if I have an average life-expectancy. I will have an even greater lifetime risk of requiring a prostate biopsy, and even if that biopsy is initially negative, I will be left worrying about a false-negative result [6]. I suspect that if PSA screening is eventually shown to decrease prostate cancer mortality in the ongoing randomized controlled trials (RCTs), and even if it proves more effective than the mortality reduction of 25% seen with regular breast-cancer screening, it would be unrealistic for me to expect to be able to reduce my risk of dying from prostate cancer from 3% without testing to much less than 2% with testing, a 33% reduction. Even if I had data from RCTs now that allowed me to be confident of this benefit, I am still not sure I would change my mind about having a PSA test at age 52, given the substantial disadvantageous risks. Therefore, I will continue to eschew screening for now, but watch the literature carefully for new developments.

I stress that I do not consider myself a nihilist in terms of early cancer detection; in fact, I decided to undergo a colonoscopy for colorectal cancer screening two birthdays ago. However, that early detection can reduce colorectal cancer mortality has been proven in RCTs. Moreover, screening for colorectal cancer reduces, rather than increases, the risk of colorectal cancer, by detecting and removing potentially pre-cancerous polyps.

If I am one of the 3% of men destined to die from prostate cancer, I may regret my decision. Although those 30-to-1 odds against are pretty long, is there anything else I can do to try to reduce my risk of dying from prostate cancer? Numerous micronutrients and dietary factors may reduce the risk from prostate cancer; the most promising, with risk reductions of up to 50% suggested in studies thus far, appear to be selenium, low-dose α-tocopherol, and even red wine (a problem with finasteride chemoprevention is that many physicians would feel compelled to order PSA tests if they prescribe this agent, negating its chemopreventive advantage) [7,8]! The benefit of trying to reduce prostate cancer mortality through these preventive manoeuvres is if that (if effective), unlike screening, they would reduce the cumulative risk of a prostate cancer diagnosis as well as death, and would therefore be dramatically less intrusive for people and more cost-effective for populations than screening. While the effect of nutritional interventions on prostate cancer incidence and mortality needs to be confirmed in large-scale RCTs, exactly the same thing must be said of PSA screening. So for now, I will wash down some vitamins and minerals with a glass of good red wine for my birthday, and revisit the decision, if good fortune allows me to do so, next year.