Much has been made of the U.S. Preventive Services Task Force (USPSTF) ”Grade D” recommendation against ”prostate-specific antigen (PSA)-based screening for prostate cancer.” [1] I’ve been logging the most recent comments from major news outlets on my Google Plus profile.
I’m in the process of dissecting the USPSTF recommendation. It’s important the public understand some of the subtleties that have created the wide-sweeping and controversial recommendation. I’m going to blog my salient findings. This blog is my first in the series.
The USPSTF sought to answer four “key” questions related to prostate cancer. The first is probably the most difficult to answer (and overall controversial).
Key Question 1. Does PSA-Based Screening Decrease Prostate Cancer-Specific or All-Cause Mortality?
First, understand the difference between cancer screening and cancer testing. A “screen” is an exam applied to everyone in a population. A good example is the colonoscopy exam. It is recommended everyone over the age of 50 years-old receives a colonoscopy exam. A screening exam, like a colonoscopy, can’t tell if a patient definitely does or doesn’t have cancer, but the screen can identify risk, and that’s the point. If gastroenterologists see a suspicious polyp during colonoscopy, they biopsy (or excise) the polyp and send it to a pathologist. The pathologist looks at the specimen under a microscope and determines if the polyp is cancer. The biopsy is the diagnostic cancer test. Everyone gets a screening colonoscopy. Not everyone has polyps, so not everyone gets a biopsy, or a specific cancer test.
Regarding prostate cancer, digital rectal exam (DRE) and prostate-specific antigen (PSA) measurement are screening tests. If either are suspicious, patients then undergo prostate biopsy, which is a cancer test. It’s possible the PSA screen saves lives by detecting cancer that would otherwise kill patients. It’s also possible the PSA screen detects cancer that is too slow growly to threaten the course of a patient’s life. If the latter is the case, patients with positive PSA screens undergo unnecessary biopsies, surgeries, and related complications.
(Again, I want to emphasize a screening test is applied to an entire population, usually at a certain age, regardless of symptoms. The traditional age for PSA screening is 40 years. The concept of routinely screening a 40 year-old man is different than screening a 39 year-old patient complaining of suspicious symptoms and family history. The urologist may perform a DRE and PSA screen in this situation, irrespective of the patient’s age.)
The USPSTF used five studies to answer whether PSA screening alone saves lives. The USPSTF classified three of the studies as “low quality,” still it included all three in the recommendation. I want to highlight one of low-quality studies.
The Quebec Trial [2]
Starting in 1988, nearly 47,000 men in Quebec were enrolled and randomized to a “screen” or “no screen” group. Men in the “screen group” underwent DRE and PSA measurement at the first clinic visit and only PSA measurement at scheduled follow-up visits. A PSA level greater than 3.0 ng/ml was threshold for an abnormal screen. Men in the “no screen” group were treated at their physician’s “usual care” discretion, not a screening schedule (e.g., the men were only screened if physicians thought their patients specifically warranted screening).
After 11 years of follow-up, the Quebec authors reported a 62% reduction in the risk of prostate cancer-specific death in screened men. There are two problems with this analysis. First, there wasn’t great follow-up in the “screen” group: only 23% of men completed all of their follow-up. Second, 7% of men in the “no screen” group were screened for prostate cancer, and their data was included in the final analysis showing reduced prostate-cancer death. So not only did nearly a quarter of the “screened” men not complete the study, but 7% of “non-screened” men were essentially plucked from (or for, depending on your perspective) the final data.
An independent organization called the Cochrane Collaboration examined the Quebec trial in 2006. [3] The data was re-analyzed according to an “intent-to-treat” principle, meaning data from men in the “screen group” were held firm against data from men in the “no screen” group. The re-analysis controls for that 7% of men in the “no screen” group that were eventually screened and moved into the “screen” analysis. Intent-to-treat analysis showed no statistically significant difference in prostate-cancer specific death between the two groups. When organized like this, the Quebec data was essentially a wash: screening patients for prostate cancer didn’t save lives.
The Quebec trial is important because it highlights some of the inherent difficulty in studying prostate cancer screening. Recall your basic science courses from high school. To prove effect in science, you have start with two identical groups, manipulate one variable, then measure the end result. In medicine, this is called a “randomized controlled trial” (RCT), and in the world of prostate cancer screening, PSA measurement is the variable (or “treatment,” which is the common term in most medical studies, even if it isn’t a true medical therapy).
It’s hard to design a true RCT for prostate cancer screening because it’s unethical to knowingly withhold “treatment” (PSA screening) that could be beneficial from a research subject (a patient). What happened to that 7% of men in the “no screen” group? What about their “usual care” prompted a PSA screen? We don’t know. I call this knowledge gap the “Quebec effect.”
Perhaps more important than stratifying patients as firmly “screen” and “no screen” is experimenting with follow-up time (e.g., do patients receive annual PSA exams? PSA every five years?) and PSA parameters (e.g., is 3.0 ng/ml, like the in Quebec trial, an optimal PSA threshold for biopsy? Where does the rate of PSA change/velocity fit into the bigger picture?).
So what do you believe? If you’re the physician, do you strongly urge your patient to forgo PSA measurement because re-analysis of studies like the Quebec trial say it doesn’t matter? Or do you hedge your bets, think about the 7% of men in the “no screen” group, and collect the patient’s PSA? Would routine screening save your patient’s life?
The USPSTF thinks the answer is a firm “No.” It wants to ignore the the Quebec Effect, but that doesn’t mean it isn’t real. Combine the “screen” group and the men who were screened, for whatever reason, in the “no screen” group, and we see effect. Are we supposed to ignore the Quebec Effect in the name of statistical analysis trying to preserve RCT cause and effect? What about the quarter of “screen” patients who didn’t complete follow-up?
Honestly, we don’t know. At least not yet. Data is still maturing. Does 11 year follow-up really matter? If a men is first screened when he is 52 and undergoes prostatectomy surgery when he’s 55. I expect him to be living when he’s 66 years-old, PSA screen or not. I want to know if he’s living at 76 years because his cancer was taken out.
The bottom line is the Quebec trial raises more questions than it answers, yet it was a major component of the USPSTF recommendation. Unfortunately, I don’t think a wide-sweeping, firm recommendation against PSA screening based (in part) on the findings of studies like the Quebec trial is helpful for patients or physicians.
References
[1] Screening for prostate cancer topic page. US Preventive Services Task Force Web site. http://www.uspreventiveservicestaskforce.org/uspstf/uspsprca.htm. Accessed March 2012.
[2] Labrie F, Candas B, Cusan L, Gomez JL, Bélanger A, Brousseau G, et al. Screening decreases prostate cancer mortality: 11-year follow-up of the 1988 Quebec prospective randomized controlled trial. Prostate. 2004;59(3):311-8.
[3] Ilic D, O’Connor D, Green S, Wilt TJ. Screening for prostate cancer. Cochrane Database of Systematic Reviews 2006, Issue 3. Art. No.: CD004720. DOI: 10.1002/14651858.CD004720.pub2.
