Why we have a prostate cancer screening controversy


Category : Medical Rants

Several days ago I wrote about the USPSTF recommendation against prostate cancer screening.  This recommendation is causing some controversy, especially with urologists – Panel’s Advice on Prostate Test Sets Up Battle.  

What should physicians do?  How should patients put this controversy into perspective?

Here is the problem:  the USPSTF has humans; the American Urological Association has humans.  While the data seem objective, interpretation always has a subjective component.  My friend and colleague, Allan Detsky, wrote about this problem persuasively – Sources of bias for authors of clinical practice guidelines.

To improve the validity of their guidelines, they need to recognize all of the other sources of bias as well. In doing so, they will face the challenge of balancing the expertise of those with intimate knowledge, who are more likely to be subject to these forms of bias, with nonexperts who may have less knowledge but fewer of the influences that contribute to bias — no easy task.

As a generalist, I view most problems focused on the patient.  I worry about side effects.  I worry about errors of commission.

In this case, urologists naturally focus on the prostate, and the management of prostate cancer.  They worry more about errors of omission.  They have a bias towards more aggressive treatment.  I have a bias towards less aggressive treatment.

In the NY Times article from Friday we read this explanation:

At the heart of its advice is the startling finding that thousands of doctors in the United States have been doing many of their patients more harm than good. While the panel did not explicitly level such a charge, Dr. LeFevre said that the dangers of common treatments were what drove the members to recommend against screening. “If you’re the guy doing the treatment, that’s pretty hard to swallow,” he said.

Sure enough, urologists — the doctors who most often treat prostate cancer — promised to fight. The American Urological Association issued a statement saying that the recommendation “will ultimately do more harm than good.” Many urologists reacted angrily.

“All of us take extraordinary issue with both the methodology and conclusion of that report,” said Dr. Deepak Kapoor, chairman and chief executive of Integrated Medical Professionals, a group that includes the nation’s largest urology practice. “We will not allow patients to die, which is what will happen if this recommendation is accepted.” He and other urologists said that the P.S.A. test is just one part of an overall strategy that, in the hands of well-trained doctors, can help prevent death and other consequences of cancer.

Treating patients with prostate cancer is a highly profitable business in the United States, and much of the practice of urology is dedicated to this fight. If men no longer get screened routinely, urologists will see a steep decline in patient visits and income. But Dr. Kapoor rejected the notion that profit plays any role in his defense of screening.

“That I’m going to treat patients that don’t need therapy is morally repugnant,” he said.

But Dr. Otis Brawley, chief medical officer of the American Cancer Society, suggested that is what doctors like Dr. Kapoor are doing. “We in medicine need to look into our soul and we need to learn the truth,” he said. “If your income is dependent on you not understanding something, it is very easy not to understand something.”

I favor the USPSTF viewpoint, but I do understand why the urologists have their viewpoint.  

The challenge in most medical situations comes because even the best treatments have untoward consequences.  We physicians have a responsibility to balance the risks and benefits, but that balance is not objective, rather it has a subjective component.  Because we view the problem differently, we likely cannot develop a consensus.  And that means that we are human beings, not computer decision aids.

Comments (28)

There's an additional bias which you did not mention, and that is the bias of patients.  The American perception of health care is that more testing is always better than less testing.  False positives, side effects, radiation dosage, cost, etc. are factors produced by over-testing which do not normally figure into the calculus.

The current PSA controversy recalls the recent brouhaha over mammography.  What needs to be recognized is that all diagnostic tests are economically inefficient if utilized in an asymptomatic population because the error rates are high enough to obscure true positives. An elevated PSA should lead to a rectal exam by a knowledgeable physician; if the PSA stays up over time, there's a place for ultrasound.  Blind prostate biopsy in the absence of a suspicious lump does not have a positive risk-reward balance.  The same is true for mammography in the absence of a palpable lump; there are no convincing data that blind biopsy saves lives.  But, patients will continue to demand biopsy because of their fear of cancer. And, the lawyers are always waiting in the wings.

[…] – Panel's Advice on Prostate Test Sets Up Battle. … Originally posted here: Why we have a prostate cancer screening controversy — db's … ← How Acai Berries Prevent Colon Cancer | […]

Regarding generalists worry about the patient, urologists worry about revenue…, the USPT worries about population statistics and healthcare economics.
Regarding side effects, many physicians seem to be unaware of proton beam therapy, which has fewer side effects than other treatments.

If you're worried about patients and worried about side effects, then do the blood test. If the PSA is elevated then repeat it. If it stays elevated, then make sure the patient understands the risks of a biopsy and let the patient decide. If the biopsy comes back positive for cancer, then make sure the patient understands the risk to his length and quality of life from that particular cancer, and make sure the patient understands the risks of treatment, and let the patient decide.
This attempt to extrapolate directly from screening to treatment outcome does yield mathematical models, but it necessarily removes individual patient characteristics from the equation. Not all prostate cancer is the same and not all patients are the same. If you stop screening to save some patients from over-treatment, those patients who need treatment will die. Isn't the counseling of patients about those individual characteristics the job of a physician?
Blood tests don't cause side effects. Get the information and do your job.
How much of this opinion comes from a mistrust of urologists, who will end up doing much of the counseling once the biopsy process has begun?

In db's two posts on PSA, you've managed to call him "irresponsible, unethical, [and] unprofessional", implied that he hasn't "gotten the information" and told him to "do [his] job". Since it's all but obvious to anyone who reads his blog that he fits none of these descriptions, the only issue is whether or not he's gotten the information. Well, if he hasn't, he seems to be in good company, since the person who discovered the PSA, Dr. Richard Ablin, has called it a "public health disaster" (http://www.nytimes.com/2010/03/10/opinion/10Ablin.html).
I don't mind you taking issue with db's or Ablin's or the American Cancer Society's or USPSTF's statements. As you note, the American Urological Association has done so as well. But if your strategy is to call people names and impugn their professionalism, it indicates the possible bankruptcy of your position.
Best, Billy

Billy, it was not my intention to offend anyone. In my opinion, a physician has an ethical responsibility to stay informed of health issues and properly counsel patients according to that information. To blend screening and treatment into one math problem, and then give patients advice based on the product of that problem while excluding other relevant medical information, would appear to me to neglect that ethical responsibility. If you can suggest gentler way to make that statement I welcome the suggestion.
I also invite you to address the central point of my post: Blood tests don't cause side effects. It is the reaction to those blood tests that can lead to complications, so let's address the decisions made after a PSA test comes back elevated. If we continue with PSA testing and then help patients make wise, fully informed decisions about when to biopsy and when to treat, then we can reduce unnecessary side effects while still saving the lives of that significant minority of men with aggressive, localized, asymptomatic prostate cancer.
If we simply discontinue the test then some men are going to die when their lives could have been saved. My suspicion is the men in that category would feel I wasn't being blunt enough.

I think one of the areas of disagreement here is that the USPSTF is coming from a population standpoint and Mr Orwig is coming from an individual standpoint. Suppose we have a test that detects cancer in a certain percentage of patients and a  small percentage of patients so discovered do better than if they were unscreened. But also suppose the treatment alters lipid levels possibly increasing the risk of coronary disease, causes osteoporosis with attendant increased risk of hip fractures and attendant comorbidities, and surgery for the condition  entails some mortality, so that there is little if any mortality gain for the screened population. For every individual saved by the screening one dies as a result of the therapy. Further suppose the treatment has lots of unpleasant side effects. One individual in the screened group might benefit from the screening but others may be harmed and the net outcome for the screened population is not any better. A task force looking at population outcomes will not support such a screening test. A medical group dealing only with the cancer might support the test.
What is clear is that we need better tests and ways of predicting the behavior of prostate cancer.

I agree, Mt Doc, and I think making individual decisions based on that population data is — I'll state politely — misguided and based on too little data.
An elevated PSA score sets off a long cascade of decisions to be made by patient and doctor. Should we do a biopsy? How aggressive does this cancer appear to be? What evidence is there of spread? Should we do a bone scan? An MRI? Should we treat it at all? How should we treat it? Once a treatment decision is made (none, radiation, surgery, others) a multitude of decisions must be made about how aggressive that treatment should be. There is much room for improvement in how these decisions are made, and some research suggests we are doing too much and doing it too aggressively. Men who would not die of prostate cancer are suffering side effects that could have been avoided.
But the USPSTF recommendation is just to scrap the whole thing. Don't screen asymptomatic patients. The problem is that process — while flawed — is saving lives. If we just scrap the whole thing and wait for symptoms then young men with aggressive prostate cancer are going to die when they could have been saved. For a practitioner it seems clear to me that is the wrong choice.

Problem is when data goes against our BELIEFS. Scott believes PSA screening saves lives. The PLCO prostate cancer screening trial begs to differ. This trial randomized 76000 men to screening or no screening. The unfortunate TRUTH is that sceening didnt save lives (the incidence of DEATH per 10,000 person-yrs was 2 in the screened group and 1.7 in the unscreened group). OOOPs that goes again what Scot and the Urologists KNOW to be true that screening saves lives. You can identify prostate cancer early but for the most part you still will die of what ever was going to kill you.
The USPSTF tries to be as unbiased as possible and to include broad groups of people. It also only looks at data. Interest groups like urologists dont care about the data when it goes against what they KNOW to be true.
Screening will always be populational. Thats what it is designed for. Cant make the argument about the individual and divorce it from the population. The population is made up of individuals. What we need to do is identify a high risk group and then develop a way to screen them. Maybe PSA is the right way in them; maybe it isnt. We need further study.
In the meantime it is clear from RCT data that screening doesnt save lives using PSA and therefor PSA should be abandoned.

 "the USPT worries about population statistics and healthcare economics"
No, USPT does not make recommendations based on healthcare economics.
" To blend screening and treatment into one math problem . . .  would appear to me to neglect that ethical responsibility."
A screening test is only valid if it leads top safe and effective treament. To SEPARATE the two is what would be unethical.

Terry, there is plenty of data showing men with early stage aggressive prostate cancer survive longer with treatment. I'll leave you to do some reading on that.
My problem with the USPSTF recommendation is its reliance on an averaging together of large classes of patients and cancers. Some of those patients need treatment. Some do not. If we don't screen we won't find the ones that do need treatment, and they will die of a treatable cancer. 

"A screening test is only valid if it leads top safe and effective treament."
The PSA test is a blood test. It does not cause side effects. It does not treat cancer. It just gives a physician and a patient a piece of information. If our problem is with poor decisions being made with that information, then lets address those decisions. If we decide instead to ignore the information altogether because we don't like what is being done with it, then men who could have been saved will die.
I keep pointing out that men with asymptomatic but aggressive prostate cancer will die unnecessarily if we don't screen. I don't think that's in doubt, is it? The USPSTF data doesn't dispute it. Why is that fact so completely unpersuasive?

Scott you have not date to support screening. Treatment and screening are 2 different things. After you screen you get treatment. Thats what happened in PLCO. It didnt work. Sorry but that is the fact. All the early stage treatment trials are just that treatment trials not screening. Well if you find it by screening and treat early then you will save lives right? Nope….PLCO proved that.
I know it seems that it should work but it doesnt.
When you read the treatment trials also pay attention to the side effects. Lots of men get impotence and incontinence. Very few are helped. Unfortunately once you open up the PSA pandoras box its hard not to go down the treatment pathway.
I for one will not get myself screened unless we develop a good test. PSA isnt a good test.

Terry, is it your position that a relatively young man with asymptomatic but aggressive prostate cancer should not be treated?

I have no problems with treatment. My own practice is to lay out the pros and cons to my patients and let them decide what they want to do about screening. My bias is against PSA screening but I let them decide. If they ask me what I would do I tell them I wouldnt get screened.

I'll repeat the question, because this is important to your patients, yourself, and your insurance company: is it your position that a relatively young man with asymptomatic, aggressive prostate cancer should not be treated?
If you think that the meta-analysis behind the USPSTF says the answer to that question is "no", then I think you are seriously over-interpreting that data.
If the answer is "yes", then the only way you're going to find those men is to do the blood test.
Given ALL the research data (not just the USPSTF analysis), it seems to me the only defensible approach is to do the PSA and then be smart about reacting to an elevated score. Hopefully we'll be able to do away with PSAs in the near future but in the mean time this approach is the only one I would want to try to defend.

The isssue is not whether a young man with prostate cancer should be treated. It is whether his outcome is any better if the diagnosis is made earlier by a screening test. If there is no difference in outcome the screening test is not helpful. Earlier diagnoses does not almost mean greater cure rates.

If the cancer is aggressive, then the outcome is improved by early detection, particularly in younger men who can be expected to live for decades before something else kills them. If an aggressive cancer has developed to the point it is creating symptoms then odds are good it has already spread. There is plenty of data to show that, but if you know about cancer and prostate anatomy you don't even need to see the data. Aggressive cancers spread, and there has to be a lot of it to create symptoms. Most heartbreaking is when the presenting symptom is bone pain from mets.
Again, most prostate cancer is of the less aggressive variety, which is why the results of the meta-analysis used by the USPSTF is so unsurprising. What is surprising is that they made such a tragic mistake in interpreting the data, and that we're even having this conversation. If we don't screen we're going to miss aggressive cancers and men are going to die when they could have been saved. Not most men with prostate cancer, but some. The group is small enough that it didn't show up when averaged with all men with prostate cancer, but big enough that you don't want to be the one to tell them they have cancer and it's too late to save them.
That fact is well established by plenty of research and by years of clinical practice. Its not just an impression and it's not a conspiracy by Big Urology. Men are going to die because of this.

I think the Task Force has misinterpreted the data, and some of these discussions are muddying things.
1. The U.S. screening study is far inferior to the European study, because in the U.S. study a very high proportion of the control group was also screened.
2. Any reasonable benefit/cost analysis of screening, including the Task Force's, is actually dominated by the treatment decision: specifically, what is the Number Needed to Treat (NNT) to reduce one prostate cancer death, and what proportion of patients experience serious side-effects?
3. Analysis of the European study shows NNT of 48 after 9 years, 18 after 12 years.
4.  All the existing screening studies have too small a sample size to have sufficient power to reliably detect plausible effects of PSA screening on OVERALL mortality. Look at the standard errors and confidence intervals. Overall mortality rates in these samples are on the order of 20%, and the expected PSA screening effect on mortality after 9 years is around 0.1%, and any effect gets lost in the overall mortality data. We have to rely on the NNT estimated from effects on prostate-cancer specific mortality, which can be estimated more precistely because prostate cancer  specific mortality in these studies is on the order of 0.8% after 10 years.
5. The fundamental issue is: do the benefits of reducing one prostate cancer death out of 48 treated (or 18 treated) outweigh the proportion of the other 47 (or 17) who have serious side-effects? This is a question that will have different answers for different people.

Thank you for adding that information, Tim. My impression (no survey data to back this up) is that the field in general regards that NNT as unacceptably high. My impression is that we are treating too many indolent cancers.
The 12 year data is probably more valid to look at given the typically slow progression of prostate cancer.
Hopefully by improving the decisions that are made in practice when a PSA comes back elevated we can drive that NNT down. Not all people who are treated have serious side effects but it appears the majority do.

I like your summary in point #5, and I tend to agree.  However, if only one of the 47 (or 17) overtreated men dies for a reason connected in some way to the overtreatment itself, you're left with a bunch of overtreated men, with no mortality benefit  to show for it.
It's not a stretch to suggest that may be the case. 

That's a good point, pcb, and is a good reason to weigh the option of no treatment carefully with every patient. There are risks beyond incontinence and impotence. But I don't know of any prostate cancer treatments with a 5% death rate, do you?

Maybe their death isn't directly related (either in diagnosis or time frame) to the  prostate cancer treatment, but the  treatment nonetheless increases risks for overall death compared to no treatment.  (infections, pneumonia, thromboembolic events, procedures, complications.  Even "softer" effects on health like stress,anxiety,depression may contribute).  That's why trying to show an overall mortality benefit, though difficult,  is important. 
I don't know if the studies can tease that out or not. 

…..addtionally, there may be a contribution to overall mortality in those not treated, but biopsied, or even just those who had to deal with having an elevated PSA (stress/anxiety).   It's hard to make the case those are completely benign or neutral events for people. 

My back of the envelope calculation is that if the overall mortality rate for a given group of men over some time period is 10%, and the "prostate screening" effect on mortality that we are trying to detect is 0.07% over that same time period (the effect estimated in the European study), then we probably would need to enroll 3.2 million men in a random assignment study to statistically detect this effect. I would need to do a more formal power analysis to be sure, but I am sure that the required sample size is much too large to ever be feasible. Overall mortality rates are too large and create too much noise to detect such small mortality effects in anything but the largest samples. 
I also should note that in the European study, the point estimate for overall mortality shows a reduction that is about what one would predict if there is zero offset for other cause mortality, that is, that the only mortality effect of the screening was to reduce prostate cancer mortality. I should note that in the European study, deaths during prostate cancer surgeries were counted as prostate cancer related mortality. However, the confidence interval for the estimated effects on overall mortality is so large that we cannot reject effects on overall mortality that are 3 times the reduction in prostate cancer deaths, on the one side, or an increase in overall mortality that is twice the reduction in prostate cancer mortality. There is simply too much noise in the overall mortality numbers, due to all the random effects that can affect all-cause mortality.

I would like to weigh in, as one of the minority of younger men (43 at the time) treated aggressively for prostate cancer. I was asymptomatic but had a long history of prostatitis. DRE was always negative but after a number of years my PSA level spiked despite all kinds of medication to reduce or eliminate other causes. The first biopsy was negative but elevated PSAs led to another biopsy a year later which unfortunately showed a tumor. Surgery did not result in a clean margin, so post adjuvant radiation was done. Now, more than seven years out my PSA is non-detectable. The side effects have not been fun, but I can live with them.
That is the point. I can live with them. As a young man with young children my first thought was survival. Did I over-treat? From my point of view, hell no. Unless there is a definitive test telling me not to worry about the malignant tumor growing down there, I'm going to do what any sane individual would do and eliminate the risk. Given my age and risk factors I am thankful that I at least was given that choice, unlike other men who found out too late after anything could be done.

You wre not an asymptomatic man being screened. You are not the population being discusssd.
There is no recomendation by any organization that I am aware of that recommends screening asymptomatic 43 year olds.

Post a comment


+ one = 3

  • Technocrati
  • Twitter