Saturday, January 5, 2013

PSA testing: excerpts from a roundtable discussion

Last fall, the editor of the Journal of Lancaster General Hospital invited me to participate in a roundtable discussion of implications of recent evidence on the prostate-specific antigen (PSA) test. An edited transcript of that discussion appears as an article in the Winter 2012 issue of the journal. Thanks to oncologist Randall Oyer, MD and urologist Paul Sieber, MD for a valuable and insightful discussion. Read the whole article, if you can; key excerpts from my side of the transcript are included below.

On the conflicting PLCO and ERSPC randomized trials of PSA-based screening:

I think a lot of people in the urology community sort of dismissed the PLCO and said it was hopelessly contaminated and we should just pay attention to the ERSPC; conversely, there might be people saying “we should only pay attention to the PLCO because it is the U.S. study.” I think both studies have things to tell us, and the truth is somewhere in between.

It’s important to emphasize that in the PLCO study, the “contamination” caused by opportunistic screening in the control group would have narrowed any apparent differences between the controls and the study group, and would have minimized any apparent benefit in the study group. Also, “contamination” would not cause what the PLCO continues to show, namely that there seems to be more harm from the screening than good. Furthermore, you would think that even if “contamination” made the benefit seem smaller than it really is, if you followed the groups long enough any real benefit would show up. But that hasn’t happened.

With the ERSPC, there are concerns about the variation among individual countries in the trial. A few had very impressive results while several of the others didn’t show a statistically significant benefit. The variation is problematic because, at least in the Goteborg Sweden portion of the ERSPC, which was published separately because its results were so impressive, men in the group assigned to screening were – for obscure reasons – more likely [than those in the control group who were diagnosed with prostate cancer] to get their treatment from university centers. That doesn’t necessarily mean they got better care, but it is a little bit problematic.


On the intersection of lead-time bias and clinical experience:

The observation that the practice of urologists has changed is obviously not something we can experience as primary care physicians, but their experience of seeing more men with treatable cancers can be misleading in that overall we are diagnosing more men with prostate cancer than we did before. Many of these are men who never would have found out they had prostate cancer. It is estimated that up to 50% of men in the ERSPC study were over-diagnosed in that their prostate cancer probably would not have presented in their lifetime. So it dilutes the groups because you are adding a lot of men who thought they were healthy, but as a result of the screening they have been diagnosed with prostate cancer. We treat the prostate cancer and feel good about it because these men do well, but maybe they would have done well anyway.

On the challenges of measuring morbidity in prostate cancer:

Morbidity is important but it is very, very hard to measure in an unbiased way in a screening trial. We actually looked for that information in 2008; we discussed a center in the ERSPC that tried to measure cancer metastases as an outcome and, in fact, showed that men in the control group were more likely to have metastatic prostate cancer than men in the screening group.

The problem is that it is hard to avoid bias when you are measuring that, because although for some men metastasis would present as bone pain, and then you would do appropriate diagnostic studies to confirm the diagnosis, for others the reason they had a metastasis discovered was because their physician knew they had prostate cancer and then ordered a bone scan and found it, though the patient didn’t have any symptoms. That would be more likely to occur in the screening group because they are the ones who are getting PSA tests, whereas in the control group you could have men walking around for some time who wouldn’t know that they had metastatic prostate cancer. So, I agree that if we could measure the morbidity it would be an important outcome, but it is very challenging to do that and be assured that we are not overweighting it on one side or the other.

On screening younger and "high risk" men (e.g., African Americans)

Young men have longer life expectancies, and theoretically would be more likely to benefit [from PSA screening], but I don’t think we know enough about the differences between the tumors that young men get versus the tumors that older men get. There was an intriguing finding in the ERSPC that men between 50 and 54 actually did not experience a decrease in mortality from prostate screening. Though that may have been an artifact of statistical power related to how many men were in that group, it does suggest that it may not be correct to automatically think that the younger the men are, the more likely they are to benefit.

We often tell African-American men and men with a family history, “you are at high risk and you should get screened.” The problem is that although they are indeed at higher risk, we are not certain that we can benefit them more by screening. Even though it sounds intuitive to say “well they are at higher risk, they should be targeted for screening," none of the trials have shown more benefit from screening in those patients than in the general population.

For physicians who don't want to abandon PSA screening:

I hope that, at a minimum, even if people don’t follow the [U.S. Preventive Services] Task Force recommendations and completely discontinue prostate screening, we will have improved the quality of discussions the patients are supposed to be having with their physicians about what their risk is, what outcomes they value, and what they are willing to endure to make sure that they don’t develop late stage prostate cancer.