Archimedes -
You make some interesting points. The assumption has always been that early detection of ovarian cancer would make a big difference in outcome. That seems pretty plausible given the big difference in prognosis between early and late stage ovarian cancer, but the recent example of early detection of neuroblastoma in children provides a caution.
Unless there are stored blood samples someplace from women that later went on to develop ovarian ca, the next step has to be to screen a large number of general-population women and see what comes up. It certainly is possible that the patterns they detected only happen in late stage disease that would normally be detectable anyhow.
The Scientist had a nice article on this study:
the-scientist.com
and here's a link to the raw data if anyone is interested:
clinicalproteomics.steem.com
Finally, here's the NY Times report on the neuroblastoma disappointment:
April 9, 2002
Test Proves Fruitless, Fueling New Debate on Cancer Screening
By GINA KOLATA
or years, it was a medical truism that the earlier cancer could be detected, the better. Most cancers would inevitably worsen if left untreated, the theory went. Spontaneous remissions were so rare as to be almost unheard of.
But last week, those assumptions were shattered, at least in the case of a childhood cancer. A screening test that looked as if it would save children from terrible deaths from a cancer of the nervous system utterly failed to fulfill its promise.
Now the story of that screening and questions about tests for adult cancers like mammography and a blood test for prostate cancer are ushering in a broader debate about cancer screening in general, with questions about what is known of the benefits and risks of tests that look for cancers in healthy people with no symptoms.
The questions are pressing, because cancer remains a leading killer. In 1998, the most recent year for which data are available, 541,519 Americans died of cancer. Men have a 43 percent chance of developing cancer in their lifetime and for women the lifetime risk is 38 percent.
And there are no easy answers. The recent mammography debates have led some experts to suggest that women and their doctors study the issues and decide for themselves, a course of action that others say is confusing.
But as the debate over mammography and over other tests, like the one for prostate cancer, burn on, the childhood cancer test is giving some experts pause.
The cancer was neuroblastoma, an attack on the nervous system that is one of the most common, and most lethal, tumors in children. Researchers in Japan had found that a urine test could find signs of the cancers long before symptoms appeared.
A large screening program started in Japan, while studies of the test began in Quebec and Germany. At first, the results looked spectacular. Many more cancers were found, and they were found early. Children underwent surgery, usually of the adrenal gland, where the tumors tend to lodge, and their cancers went away.
But to the investigators' shock, there was no decline in the number of toddlers who developed advanced cancers, and the death rates from the disease stayed unchanged.
"It was an absolutely stone cold negative outcome," said Dr. Steven Goodman, an associate professor of pediatrics and epidemiology at the Johns Hopkins School of Medicine, who was not associated with the study. The scientists had to declare that the screening test, seemingly so promising, should be abandoned.
Their finding thrust them into the debate over a growing assortment of screening tests used to search for disease in adults and children.
In an editorial with the neuroblastoma papers in The New England Journal of Medicine last week, Dr. George Cunningham of the California Department of Health Services wrote:
"As these studies illustrate, the decision about whether or not to screen should be driven not by the availability of a laboratory screening test, but by careful analysis of outcomes, including saving the lives of the screened newborns or improving the quality of their lives."
The screening debate is far from settled.
Although he agrees with Dr. Cunningham on the need to evaluate tests, Dr. Larry Norton, a specialist on breast cancer at the Memorial Sloan-Kettering Cancer Center in New York and the president of the American Society for Clinical Oncology, said he still strongly believed that early diagnosis made biological sense.
The theory behind early diagnosis, Dr. Norton added, comes from years of work with animals that repeatedly shows that cancers treated early can be cured and that if they are left to grow and spread, cure becomes impossible.
"On the basis of animal models, my predisposition is to believe that early diagnosis is a good thing," Dr. Norton said.
He said he knew of no exceptions in the animal research. There is no reason, he said, to think that humans are inherently different from animals. "Early diagnosis works in animals, and basically mammals are mammals," Dr. Norton added.
The question, he said, is whether the screening tests are detecting cancers. "Early diagnosis makes sense if you are diagnosing a true cancer," he added.
In neuroblastoma, he said, the problem was that the test was not finding true cancers. It was finding tumors that did not need to be treated and was missing those that were deadly.
To everyone's surprise, it turned out that the supposedly rare spontaneous regressions were actually the most common form of neuroblastoma tumors. No one had noticed them before because, without the screening test, very few of those tumors had ever been detected.
The Nature of Cancer
Lazy Cells Upset Traditional Ideas
Dr. William Woods, a neuroblastoma expert at the Aflac Cancer Center of Emory University in Atlanta, cited a powerful lesson: "The theory that one can pick up all cancers early and have the outcome good is not correct," he said. "It certainly is not correct for neuroblastoma."
But some are asking whether the neuroblastoma case is the exception or the rule. That is a conundrum that hinges on an emerging and startling view of the very nature of cancer.
Over the last few years, investigators have found more and more hints that cancer is not what they always thought it was. The traditional view was that cancerous cells looked aberrant under a microscope, grew in an uncontrolled fashion and, in most cases, killed.
Now, though, researchers understand that some cancers are indolent — so indolent, in fact, that they will never grow large enough in the patient's lifetime to cause medical problems. Still others look like cancers under the microscope and have growth patterns in the lab that are typical of deadly tumors but can stop growing on their own and revert to normal tissue.
At the same time, there are more screening tests, and they are increasingly sensitive, finding cancers at earlier stages. That leaves doctors with a problem. The earlier they detect cells that look cancerous, the less certain they can be of how dangerous they are.
"The whole issue of diagnosis is based on visual criteria," said Dr. Evan R. Farmer, a dermatologist, skin pathologist and the dean of Eastern Virginia Medical School. "But with cancer, you are talking about what happens biologically. If it metastasizes and kills the patient, then you know it's cancer. If it doesn't metastasize, then you don't know if it is or isn't cancer.
"Maybe you didn't wait long enough or maybe you treated it successfully or maybe it had the capacity to metastasize but you removed it or maybe it looked like cancer and you called it cancer. But in fact it didn't have the ability to metastasize."
Medical experts say they are agonizing over the screening questions.
"This is both an intellectual and an emotional debate," said Dr. Isra Levy, a cancer specialist at the Canadian Medical Association, who participated in the study of neuroblastoma screening. "The issues are all legitimate issues, and they are very, very difficult. We are dealing in a world where certainty is desirable, but it isn't there."
Even the names given to some tumors hint at their ambiguous nature. There are "incidentalomas," adrenal gland tumors of unknown significance that began turning up in increasing numbers when people had C.T. scans or M.R.I.'s for other conditions. Medical specialists do not know whether they should be removed or left alone, but they do know that lethal adrenal cancers are far too rare for many of those tiny tumors to be dangerous.
In cervical cancer, the most common abnormality found with Pap tests is called "atypical squamous cells of unknown significance."
"The name tells you how much we know about its natural history," Dr. Barnett Kramer, director of the Office of Disease Prevention at the National Institutes of Health, said. The condition's significance is unclear, but to be safe, doctors remove the abnormal cells.
Experts may not even agree about whether a cell sample looks abnormal, as Dr. Farmer learned a few years ago. He and his colleagues asked expert pathologists to submit what they considered classic slides of skin tumors that were either malignant or not, examples that these specialists thought were so clear that they could be used in a textbook. Dr. Farmer and his team sent the slides of cells from 37 patients to eight leading pathologists. Those experts had agreed to decide whether the tumors were cancerous or normal or whether the category was unclear.
They looked for a disordered growth pattern in the tumor tissue. They examined the individual cells, looking for irregular shapes and enlarged nuclei. In just 11 of the 37 tumors did all the experts come to the same conclusion about what they were seeing.
The Skewed Screenings
`Overdiagnosis': A Growing Problem
Neuroblastoma is the sole cancer in which such clear evidence exists that a screening test did harm without doing good. But some experts say there is reason to believe that other cancer screening tests produce similar effects, but perhaps not as pronounced.
Those experts call the problem overdiagnosis, or finding tumors that are not dangerous but that cannot be distinguished from ones that may become lethal. Those experts see signs of overdiagnosis in prostate cancer, where men 50 and older are urged to have a blood test known as the P.S.A. (or prostate specific antigen); in melanoma screening, which looks for the malignant skin cancer; and in lung cancer screening, with a new test, the spiral C.T.
Overdiagnosis may also be occurring with screening for breast, cervical and ovarian cancers.
"Overdiagnosis exists in virtually every cancer," said Dr. Otis Brawley, a professor of medical oncology and epidemiology at the Winship Cancer Institute of Emory University in Atlanta.
Unless it can be proved that a screening test reduces the death rate of a cancer, researchers like Dr. Brawley, Dr. Kramer and others say, it may be better to forgo it and wait for symptoms than to have a test. "It's the iceberg issue," Dr. Kramer said. "If you dip below the waterline to lesions you've never seen before, you can't assume that because they look like tumors we've seen before that they will have the same natural history."
Dr. Kramer cites data from Japan on a new test to find early lung cancers. The test, spiral C.T. scanning, is finding similar numbers of cancers in people who never smoked as in smokers. Yet 10 times as many smokers die from lung cancer. That, Dr. Kramer said, is a hint that the new test, like the older ones, is finding cancers that are not dangerous.
Prostate cancer screening also raises overdiagnosis questions, some researchers say. With the widespread use of the P.S.A. test, the incidence soared, from 143.3 cases per 100,000 in 1990, hitting a peak at 195.6 in 1992 and declining to 155.3 per 100,000 in 1998. By 1998, so many men had already been tested that fewer new cancers were being found. But the death rate from prostate cancer, which fell from 38.6 per 100,000 in 1990 to 32.3 per 100,000 in 1998, did not fall nearly so fast and at least part of the decline is due to improved treatment.
"I'm very very worried," Dr. Brawley said.
Some experts, like Dr. Ian M. Thompson Jr., a urologist at the University of Texas Health Science Center in San Antonio, say that although some overdiagnosis of prostate cancer undoubtedly occurs the test is doing what it is supposed to do, finding cancers early and saving lives. He is heartened by the fact that the death rate is lower and says he sees more men with cancer that can be treated.
Dr. Thompson has the P.S.A. test himself.
Dr. Eric J. Feuer, a statistician at the National Cancer Institute, cautioned that it could be difficult to interpret national cancer data because many factors influenced the figures. In breast cancer, the incidence has risen consistently and steadily, independent of mammography, about 1 percent a year.
In fact, Dr. Feuer said, the increase apparently began in the 1940's, long before mammography was introduced. But it complicates attempts to interpret trends in incidence rates.
Attempts to interpret mortality rates are complicated by the fact that cancer treatment has been changing and improving over the years.
Dr. Norton said it would be a mistake to write off cancer screening in general because of fears of overdiagnosis. For now, he said, early diagnosis may be the only way to cure some cancers.
Dr. Norton said that with mammography, at least, the test was finding tumors so early that the combination of early diagnosis and improved treatment had led to a 10 percent drop in the death rate since the early 90's.
"The package of screening and treatment is having an impact on mortality," he said. While overdiagnosis may occur, Dr. Norton said, it may be balanced by better outcomes for women whose cancers are found at a stage when the treatment is more effective and less disfiguring and debilitating.
The Evaluations
Should Every Test Be Questioned?
Mammography, which is being subjected to vigorous debate, is one of the only screening tests that has been evaluated in clinical trials.
The debate hinges on the quality of the studies that show regular screening mammograms reduce the death rate from breast cancer. Some critics say that the benefit is small, that it takes tens of thousands of women being screened for years to find a savings of a few lives and that with such a small benefit uncertainties about the study's data make them question whether the results are solid. Other experts say the studies show that mammograms can prevent deaths from breast cancer.
But even some who defend the mammography trials say the screening questions have to be raised anew with each test.
"You want to apply the question to every cancer screen," said Dr. Robert A. Smith, director of the division of cancer screening at the American Cancer Society. "You want to be sure that whatever promise it appears early diagnosis holds, you haven't been completely blindsided."
Dr. David Freedman, a statistician at the University of California at Berkeley, said he was also satisfied that the benefits of mammography were demonstrated, but agreed that every screening test had to be evaluated. "You cannot say a priori by pure reasoning that screening helps for other diseases," Dr. Freedman said.
Dr. Bradley Efron, a professor of statistics and health research and policy at Stanford, faults experts who have oversimplified the case.
"The people who say, `We're saving lives,' are undoubtedly right," Dr. Efron said. "They probably are saving lives. But the question is, What's the cost? We could save even more lives if all men had their prostates removed at age 50 and if all women had their breasts removed at age 50."
If the number of lives saved is minuscule compared with the number of people who are harmed by overdiagnosis, by unnecessary treatments, by treatments that could have been delayed with no ill effects, then the value of the test has to be in question, he said.
"It's quite a sophisticated point," Dr. Efron said. "I don't think you can blame the public. I blame the medical researchers who say, `We're saving lives.' "
The questions are pressing because screening is different from any other medical program, said Dr. Goodman of Johns Hopkins. "We bring it to healthy people, and those who test positive become sick people, the subjects of medical intervention," he explained. "We need to be awfully careful."
Peter |
