September 17, 2002
Dilemma on Prostate Cancer Treatment Splits Experts
By GINA KOLATA
You're a man, 65 years old, and you've been having trouble urinating. Your doctor tells you that you have a cancer in your prostate making the gland press on the tube that carries urine. The good news is that the cancer seems confined to your prostate; there is no evidence that it has spread.
Now you must make a decision. You can have an operation that removes your prostate. It may leave you impotent or incontinent. You may need to wear adult diapers. But a study from Sweden published last week in The New England Journal of Medicine finds that you will also reduce your risk of dying of prostate cancer by 50 percent.
Do you want the surgery?
Your doctor may then tell you the rest of the story: the study found that after an average of 6.2 years of follow-up there was no change in the overall death rate of men in the study. The number of men who died in those years was the same in the group that had the surgery as it was in the group that did not. Why should you care about the overall death rate? After all, your problem was prostate cancer. If the death rate from it was reduced, isn't that good?
It depends, some medical experts said, and the question of how much, if at all, you should care is at the heart of a debate.
The treatment, for instance, can lead to deaths from other things.One case might go like this: Say you had the operation, but it caused blood clots. Instead of dying of prostate cancer, you died of a stroke.
At the same time, you are told that, while the surgery reduces your chance of dying of prostate cancer, the reduction is small, not enough to have an effect on the overall rate of death from all causes, including prostate cancer.
What about that 50 percent decrease in the prostate cancer death rate? The actual numbers showed that 16 of 347 men who had the surgery died of prostate cancer, or 4.6 percent, compared with 31 of 348 who did not have the surgery, or 8.9 percent.
Now, knowing that your likelihood of dying in the next six years is just the same, whether or not you have the surgery, but your likelihood of spending those six years impotent and incontinent is increased, do you still want the operation?
The dilemma is not just medical sophistry. Instead, it touches on a fundamental debate among doctors and researchers. What should be the bottom line? Is it enough to show that you can reduce the likelihood that a person will die of a particular disease? Or must you insist that the overall mortality rate is reduced?
The prostate cancer study reveals splits among medical experts. Some, like Dr. Patrick Walsh, a urologic surgeon at Johns Hopkins Hospital, say it is enough to prevent deaths from that disease.
"Have you ever seen anyone die from prostate cancer?" he asks. "It's a terrible death." The cancer, he said, moves into the bones. "Bones break and disintegrate. You linger for a year of a painful death. I have three uncles who died of it. One was a favorite uncle, and I was a child watching him die. It was a horrible thing."
Dr. Albert Mulley, a specialist in internal medicine at Massachusetts General Hospital, has a different perspective. "I see people who die of esophageal cancer or liver cancer or lung cancer," he said. "There is nothing uniquely terrible about prostate cancer."
"Is it worse than Alzheimer's?" he asked. "I don't think so. Is it worse than a debilitating stroke? I don't think so?"
The problem with demanding that treatments or screening tests reduce the total death rate, however, is that it is a high hurdle, sometimes an impossible one, researchers say.
Statistics can be against you if a disease causes proportionately few deaths. Men in their 70's and 80's, for example, are at risk of dying of many diseases besides prostate cancer, and so even if prostate surgery saved a few lives, that effect could be lost because there are so many other deaths from other causes.
It is easier to find effects on the death rate when all you look at is the deaths from the particular disease, but there are trade-offs.
One problem, said Dr. Steven Goodman, a biostatistician at Johns Hopkins, is that you may be missing risks.
"You are always afraid in the back of your mind that maybe the disease-specific benefit is being bought at the expense of raising your risk of death from other causes." He explained: a test finds a cancerous tumor and the patient has chemotherapy or radiation. Years later, the person dies of heart disease, which was actually caused by the cancer treatments, which hurt the heart. If the person had not been tested for cancer and treated, he would have died at the same time of cancer. In this case, with the early detection and treatment, "you're not better off," Dr. Goodman said.
Another problem involves the statistics: it can be hard to agree on a cause of death, and without realizing it, researchers may tilt their analysis in favor of the treatment for the disease they are studying.
"If a patient has advanced prostate cancer and then comes into a hospital and dies of pneumonia, that might be written down as a death due to pneumonia," said Dr. Timothy Wilt of the Veterans Affairs Center for Chronic Disease Outcomes Research in Minneapolis. Yet, he said, the man might have gotten pneumonia because he was weakened by treatment for his prostate cancer.
Dr. Wilt noted that in the recent Swedish study of prostate cancer, a committee reviewed all deaths that occurred. Initially, they agreed on the causes of death for 85 percent of the 115 men who died in the course of the study, eventually reaching 100 percent agreement.
"That's in a highly controlled setting where they are spending time and money to be as rigorous as possible," Dr. Wilt said, adding that most studies did not reach that level of rigor. "If they had misclassified just three or four deaths, there would have been no difference in the death rate from prostate cancer."
The rules for deciding what counts as a cancer death are rarely spelled out, said Dr. H. Gilbert Welch of the V.A. Medical Center in White River Junction, Vt. "We are not all clear on what constitutes a cancer death," he said.
Often, Dr. Goodman said, the decision to accept a reduction in the death rate from a particular disease rather than a reduction in the overall death rate comes down to money and resources. "Usually, the number of people you need to study the disease-specific effect is much smaller than the number you need to look at the overall effects," he said. "A lot of people would say, `Wait a minute, if I can show I can decrease prostate cancer mortality with 3,000 people, why do I have to go out and get 12,000 people to show the other?' Often, they don't have the choice. They don't have the money."
In the case of prostate cancer, Dr. Wilt is directing a study that is designed to find a decrease in the overall death rate, if it exists. The study involves 731 men whose cancers were found early, with a P.S.A. screening test and who were randomly assigned to have their prostates removed or not. After 12 years, the researchers expect to see an effect if there is one.
"We felt that if we haven't been able to show an increase in survival, then you have to wonder about the importance of the treatment," Dr. Wilt said.
Dr. Wilt's study, of course, is the ideal. "Everyone would say, boy, that's what I really wish I had," Dr. Goodman said. "The greater debate is, What do you do in the absence of such data?"
At the very least, Dr. Welch said, "if it's such a small difference that you can't see it in the overall mortality, then patients ought to know that. If it's that close a call, maybe they will make a different decision."
Or maybe not.
Dr. Goodman recently sat in on a consultation with a leukemia patient whose doctor told her that the treatment, a bone marrow transplant, would reduce her chance of dying from the disease. But the treatment itself could prove fatal. In the end, treatment did not budge the mortality rate.
The woman, Dr. Goodman said, chose to be treated. She said, Dr. Goodman recalled, "I'd rather go down fighting."