Saturday, March 21, 2009

PSA Screening: What is the value?

Two studies published in the New England Journal of Medicine on line on March 18, 2009 regarding the use of prostate-specific antigen (PSA) screening for prostate cancer have been getting a lot of coverage in the popular media, including NPR and the New York Times. The reason is that these studies do not, overall, indicate that such screening saves significant numbers of lives. In the US study of 77,000 men, the PLCO trial, there was no significant difference in mortality in the group receiving PSA screening (92 deaths in the study group vs. 82 in the control group),[1] while in the European study of 182,000 men, there was a very small reduction in mortality, barely achieving statistical significance.[2] Dr. Allan Brett, summarizing the article in “Journal Watch: General Medicine” notes that “To prevent one prostate cancer death, 1410 men had to be screened, and 48 additional cases of prostate cancer had to be diagnosed and treated. All-cause mortality did not differ in the two groups”.[3]

In fact, the US Preventive Services Task Force, the committee that evaluates prevention strategies including screening tests, has long indicated that there is insufficient evidence to recommend for or against prostate cancer screening using PSA, and recently amended that statement to recommend against screening in men over 75:

The USPSTF concludes that the current evidence is insufficient to assess the balance of benefits and harms of prostate cancer screening in men younger than age 75 years..
The USPSTF recommends against screening for prostate cancer in men age 75 years or older.”

So what is the fuss? Why does the Times have a front page article entitled “Studies show prostate test saves few lives”.[5] Well, mainly because, despite the lack of recommendation by the USPSTF, doctors have been ordering PSA tests for men over 50 for years. The American Urological Association has recommended it: Regarding early detection of prostate cancer, the expert panel concluded that routine PSA testing should be offered in men when:

“Age is 50 years and greater, unless the man has increased risk factors, such as genetic predisposition via family history or African-American racial status. In these "at risk" men, PSA testing should be offered between 40 - 50 years of age.”[6]

That is to say, there were conflicting recommendations. Perhaps unsurprisingly, the urologists, who are the people who see the prostate cancer and operate on it, had a different sense of both the prevalence of the disease and its bad outcomes, and their ability to alter those outcomes through intervention. But mainly, it is the issue of looking at intermediate variables rather than the outcomes of significance. In this case, looking at whether the PSA can diagnose prostate cancer, rather than whether diagnosing (by any means) and treating prostate cancer saves lives.

What? How can finding a cancer early, and treating it, not save lives? “Why,” you might well say, “I have a friend who had a routine PSA screening, and it was high. He had a biopsy, they found cancer, they operated on him, and now he is alive. Isn’t that a good thing?” Well, it’s certainly good that he’s alive. The question is: Would he not be alive, or would he be suffering the pain of metastatic prostate cancer, if he hadn’t had the cancer found and treated? And that is the question we do not know the answer to. And that is why the recent studies are so important.

Simplistically, there are two kinds of prostate cancer: the bad kind that will kill you after probably creating very painful metastases to bone, and the kind which will be lying indolently and asymptomatically in your prostate when you die of something else. But we have no way of distinguishing between these by any of our diagnostic tests. And, more important, we have no idea whether treatment makes a difference – that is, whether the people who die without treatment would have died even if treated, and if the people who survive with treatment would have survived even if not treated.

This is a really big thing. Why would we want to screen lots of men to find a relatively few cancers that we can treat when we don’t even know if treatment makes a difference in death? In the US study, it didn’t. In the European study, it made a small difference in death from prostate cancer, but it took screening 1410 men and treating 48 to save one life from prostate cancer. And the overall death rate, from all causes, was no different.

Better safe than sorry? Maybe, but what the studies did not look at was the side effects from treatment. What about those other 47 men who were treated and whose lives were not saved? The fact is that, while there are many treatments for prostate cancer, all have a significant “side effects”. Treatments such as radical prostatectomy, external beam radiation, radioactive “seed” implants, and newer procedures such as “green laser” very frequently lead to varying degrees of impotence and incontinence, not minor inconveniences! Hormone therapy, whether by removal of testosterone production by removal of the testicles (orchiectomy) or by administration of anti-androgens, have other effects including “cosmetic” ones such as feminization (weight redistribution, breast enlargement) and serious weight gain that can lead to greater morbidity and mortality. So treatment is scarcely benign. And, because we should not forget cost, the cost of the massive screening by PSA, followed by the further testing and procedures and treatment that follow it, is enormous. These studies demonstrate that even without considering cost-effectiveness or the morbidity of treatment, there is little or no benefit to screening in terms of lives saved.

However attractive the idea of screening and early detection is, both to public-health focused primary care physicians like me and to people in general, screening is only of value if it can not only identify disease in the pre-symptomatic phase, but if there is effective treatment that has a patient-important outcome: lower mortality or greater quality of life. PSA screening, at this point, does not meet this criterion.

The NPR segment covering this issue ended with a primary care physician emphasizing that the answer was not just looking at a single PSA value, but rather looking at the change over time. He noted that the ability to do this, as well as to complement monitoring the trend in PSA with serial rectal examinations, was one of the strengths of the primary care relationship. But, with all due respect, and with the great respect I have for primary care, this misses the point. Such a relationship, with its ability to monitor trends, may increase the likelihood of an accurate diagnosis of prostate cancer (i.e., a PSA of 10 that does not change over time may be less likely to mean cancer than one that goes from 2 to 4 to 6) but this still considers a positive outcome to be an accurate diagnosis rather than a decrease in mortality. Absent our ability to distinguish between “good” and “bad” prostate cancer and to know that treatment makes a difference in patient-important outcomes, greater accuracy in diagnosis may just lead to greater cost and greater morbidity.

[1] Andriole GL et al. Mortality results from a randomized prostate-cancer screening trial. N Engl J Med 2009 Mar 26; 360:1310. published on line at
[2] Schröder FH et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med 2009 Mar 26; 360:1320. published on line at
[3] Brett, A, Journal Watch General Medicine March 18, 2009
[5] New York Times, Thursday, March 19, 2009, p.1

No comments:

Post a Comment