Accueil  |  Prestations  |  Notre réseau  |  Notre équipe  |  Info Patients  |  Info médecins  |  Nouveautés

Faut-il faire une mammographie : oui, non, peut-être ?

Base de réflexion pour les femmes (et les autres aussi !)

Référence: http://www.breastcancer.org/ren_report_2002_02.html

Quelques notions importantes issues de ce texte

What these researchers forgot to do is to ask the women themselves how they perceive the relative benefits and side effects of mammography.

Follow existing screening recommendations: annual mammography after age 40.

Based on your own particular situation, your doctor may suggest that you start screening earlier.

Limitations of mammography : the probability of missing a breast lesion is only 8-10%. This is the reason why any screening type of examination should not only include mammography, but also breast physical exam, to detect the 10% of breast cancers that do not appear on mammography.

The radiation exposure due to mammography is generally accepted to be negligible and should not be a deterrent for women to obtain yearly mammography.

Here's how to make mammography work best for you:

• Get your mammogram done in the best place possible.
• Take advantage of other breast imaging studies as needed to further check a possible breast abnormality seen on the mammogram.
• Choose a doctor who has experience with breast health, to minimize the number of necessary biopsies.
• Take advantage of the latest core biopsy techniques that require only tiny incisions and are therefore less disfiguring.
• Visit breastcancer.org regularly to learn about breaking news on promising new imaging studies.

Mammograms: Yes, No, or Maybe?

Understanding the Science Behind the Sensation, February, 2002

Background

Everyone's been talking about mammograms recently. Are they effective? Do they save lives? Should current guidelines be changed?

Every expert seems to have a different opinion, and it's very difficult to figure out what's really going on.

The breastcancer.org February Research News is dedicated to the mammogram issue. We'll explain the research studies behind all the recent media hype, and give you tools to understand what the experts are actually arguing about.

By providing accurate, understandable information about the recent mammogram debate, we we can help women make more informed choices about their health.
If you're interested in our position on mammograms, but not in the details of the mammogram debate, read the Letter to the New York Times that breastcancer.org signed along with 19 other breast cancer organizations, and The Wall Street Journal article in which we are quoted.

Basic Facts

A few basic things about the mammogram debate have not been stated clearly in the media. Even if you're not interested in all the research details, you should be aware of these important points:

1. No new research studies on mammography have been conducted recently.
2. The current debate is about whether or not PAST studies on mammography are reliable.
3. It is very difficult to determine the reliability of past studies.
4. The quality of mammography has significantly improved since these studies were conducted.
5. The results of research studies can be interpreted in different ways.

What all this means is that it is very, very difficult to look at past studies on mammograms and decide whether or not those studies show that mammograms done TODAY have a clear benefit.

Why Now?

So why is everyone talking about this now? Haven't the mammogram studies been around long enough for experts to check their reliability?

The answer to these questions is a long story. The story starts with an article published in the January 8, 2000 issue of the respected British medical journal The Lancet and has yet to end. At times, it may sound like a soap opera, or a bad episode of "ER." But this is how the story goes.

Prologue: It All Started When...

Two Danish researchers, Ole Olsen and Peter Gotzsche, noted a surprising finding: A 1999 study found no decrease in breast cancer deaths in Sweden, where mammograms to screen women for breast cancer had been recommended since 1985.

This finding was surprising because it suggested that after 14 years of using mammograms to screen women for breast cancer, there was no decrease in the risk of dying from breast cancer. Studies on mammograms suggested that finding breast cancer early with mammograms could save women's lives. But this did not seem to be the case in Sweden. Here are some possible explanations for these surprising results:

1. Even though mammography was recommended, not enough women were screened to significantly affect the death rate from breast cancer.
2. Mammograms in the first years of the 1999 study weren't good enough to reduce breast cancer deaths significantly.
3. The 1999 study was flawed.
4. The studies suggesting that early detection with mammograms would lead to a decrease in the number of breast cancer deaths were flawed.

Olsen and Gotzsche focused on the last possibility. They wanted to check the reliability of the screening mammogram studies that predicted a decrease in deaths from breast cancer. They found eight large studies on mammograms from different countries, including Sweden, the United States, and Scotland. All together the studies included half a million women. There was also a meta-analysis of mammogram studies from Sweden showing a 29% decrease in deaths from breast cancer. [top]

What Is a Meta-analysis?

A meta-analysis is a way of combining the results of a group of similar studies on the same subject. This allows researchers to combine many small studies into one bigger one.

Advantage of meta-analysis: In order to get reliable results, clinical trials need lots of participants. The more people are in a study, the more trustworthy the results are considered. By combining the numbers from many small studies, researchers can get results for larger numbers of people. The combined results may be more solid than the results from each individual study.

For example, suppose there are 10 different studies on mammograms and breast cancer, with 100 women participating in each—50 in the treatment group (the group being screened with mammograms) and 50 in the control group (the group not being screened). The meta-analysis will combine all 10 studies. So it will have a total of 1,000 women (10 studies times 100 women), with 500 in the treatment group and 500 in the control group. The meta-analysis will also combine the results of all 10 studies into one set of results.

Disadvantage of meta-analysis

No two studies are exactly alike. Two studies might ask the same basic question (Do mammograms reduce the rate of deaths from breast cancer?) and use the same basic methods (comparing a group of women who are screened with mammograms and a group of women who aren't screened). But they can differ in many smaller, but no less important, details. This means that results from a combination of several small studies are not as reliable as results from one large well-designed study.

For example, one mammogram study might screen the 50 women in the treatment group over a five-year period and then track all 100 women for 10 years after the screening is over. Another study might screen the treatment group for two years and then track all 100 participants for five years after screening started.
If, from the very beginning, one of the two studies had included 200 women instead of 100, would that study have the same results as a meta-analysis that combines the two studies of 100 women? Not necessarily.

Act 1: Researchers Claim No Mammography Benefit

Article published in The Lancet, January 8, 2000
In their 2000 Lancet study, Olsen and Gotzsche did two things:

1. They checked the eight studies on mammograms and the Swedish meta-analysis to see if the results were trustworthy.
2. They did their own meta-analysis of the mammogram studies. To check the reliability of the mammogram studies, Olsen and Gotzsche looked at the methods used to put together each study. They focused especially on how participants were selected and then divided into randomized groups.

What Is Randomization?

For a clinical trial to reveal unbiased and meaningful results, the groups of participants must be similar and balanced.

For example, to see if a medication causes more hot flashes than a placebo (a sugar pill that looks like the real medication), researchers must make sure that the group of women taking the medication and the group taking the placebo have the same number of pre-menopausal and post-menopausal women. Otherwise, if the group taking the medication has more pre-menopausal women and they have fewer hot flashes, there'll be no way to know whether it's because of the medication or because they haven't gone through menopause.

In large studies, if researchers assign people randomly to the different groups being studied (without trying to balance the groups), chances are that they WILL end up with a good balance. In small studies it's more difficult to make sure that the groups are balanced.

To be balanced, the groups should be as close to identical as possible in terms of many different factors that could affect the outcome of the study. These factors usually include age, previous history of disease, risk factors for the disease being studied, family history of the disease, social situation, financial circumstances, and level of education. Many other factors may also be considered.

Why is this "balancing" important? Because if there are big differences between the two groups, it's difficult to say whether or not those differences are affecting the results of the study.

For example, if researchers want to conduct a randomized study on mammograms involving 100 women, they can't just divide the 100 into two equal groups. They have to make sure the women in the two groups are of similar ages. If one group has more young women and one group has more older women, the results of the study might be influenced by age, which is not what the study is supposed to be looking at.

Olsen and Gotzsche found that of the eight studies, in their opinion only two were "properly" randomized. They thought that the groups being compared in the other six studies were "poorly » randomized because the participants were not similar enough in terms of age and other important factors.

Then they did a meta-analysis, comparing the results of the two "properly" randomized studies to results of the other six. They found that in the two, there was no significant difference in the risk of death from breast cancer between groups of women who had been screened with mammograms and groups of women who hadn't been screened. In contrast, the combined results of the six "poorly" randomized studies showed a significant decrease (about 25%) in breast cancer deaths.

Olsen and Gotzsche did not do a meta-analysis of all eight mammogram studies together (which would have shown an overall decrease in breast cancer deaths in the screened group) because they did not want to include "poorly" randomized trials with "properly" randomized ones.

They concluded that the differences in the findings between the two groups of studies—properly randomized and poorly randomized—were due to problems with the second group of studies. They supported their claim by citing research which shows that studies with poor randomization the effect of the treatment (or screening method) being tested by 33-41%.

In other words, Olsen and Gotzsche concluded that the only reliable studies showed no benefit for mammography in reducing breast cancer deaths. And they said that the six studies that showed a large advantage to mammography did so because they were poorly randomized, not because mammography actually reduced breast cancer deaths.

Act 2: Same Researchers, Same Claims, Same Evidence

Article published in The Lancet, October 20, 2001

The 2000 Lancet study understandably received a lot of attention, and some people criticized the way Olsen and Gotzsche had done their meta-analysis. So Olsen and Gotzsche went back to the eight mammogram studies and used a more rigorous method, called a "Cochrane review," for their meta-analysis.

In The Lancet, on October 21, 2001, the researchers reported that their Cochrane review confirmed their previous findings: There is no reliable evidence that mammograms reduce deaths from breast cancer. They also added that they had data showing that screening with mammograms leads to more aggressive treatment (more surgery, chemotherapy, etc.), more psychological stress, and higher costs if anything is found on the mammogram.

Act 3: Things Get Complicated

Different conclusions in different publications

Olsen and Gotzsche submitted their renewed findings to the Cochrane Breast Cancer Group for publication in the Cochrane Library. This is where Cochrane reviews that are accepted by the Cochrane Group are published. But the Cochrane Breast Cancer Group editors would not publish Olsen and Gotzsche's review unless the authors agreed to make some changes. And Olsen and Gotzsche refused.

So the Cochrane Library published one version (abstract available for free and full article available for a charge here) and The Lancet published its own commentary and commentary from the researchers (the October 20, 2001 article). The Lancet web site also published the authors' unedited version of their Cochrane review (breastcancer.org has been unable to locate that review).

In the Cochrane Library version, the editors concluded: "The currently available reliable evidence has not shown a survival benefit of mass screening for breast cancer. However, the trials in this review are still underpowered for all-cause mortality; the confidence intervals include both a plausible worthwhile and a possible detrimental effect." Translation: The Cochrane editors are saying that based on available evidence, a person could conclude either that mammograms save lives or that they are harmful.

How could the same evidence lead to two opposite conclusions? It Is possible for different researchers to look at the same studies and arrive at different conclusions. Some researchers look at mammography studies and ask only: "Do mammograms save lives that otherwise would have been lost to breast cancer?" Other researchers want to know more: "Do the benefits of mammography outweigh its disadvantages?"

Advantages, Disadvantages and Limitations of Mammography

Mammogram Advantages

No better method:
There's no other proven screening test for detecting breast cancer early.

Easier to beat smaller tumors:
Early detection provides more opportunity to discover a tumor when it is very small. This means less cancer in the body has to be eliminated. Faced with a smaller "enemy," you stand a better chance of winning the battle, using effective treatments.

Less disfiguring treatment:
Early detection means you might have more treatment options. For example, early detection may permit breast-preserving surgery by finding the tumor when it is small. If the tumor is found only when it is large, mastectomy may be required.

Less toxic treatment:
Early detection also means a lower risk of lymph node involvement. The amount of chemotherapy called for when lymph nodes are clear of cancer is usually less than when lymph nodes are involved.

Increased effectiveness when combined with other tests:
Mammography is only the first in a series of tests that can be done if something looks abnormal. If a mammogram finds anything abnormal, additional tests (for example, ultrasound and MRI) can help figure out if the abnormality is cancerous or not.

Active participation in health care:
By having regular mammograms, you're doing the best you can with what is currently available today, taking an active role in maintaining your health. [top]

Mammogram Disadvantages

False positives:
A false positive is the name for a breast abnormality that at first seems to be a cancer but really isn't. False positives cause possibly unnecessary surgery, anxiety, and cost. According to a study quoted in the Cochrane Library version of Olsen and Gotzsche's review, the cumulative risk This means that if you've had 10 mammograms, you have a 50-50 chance of having a false positive in one of them.

Unnecessary procedures:
The likelihood of having a biopsy (an operation to take out a growth and check it for cancer) after 10 mammograms is 18.6%. But only a small percent of those find a cancer. This means that most biopsies following mammograms find nothing..

Unnecessary treatment for slow-growing tumors:
A mammogram may discover a tumor that is cancerous but very slow-growing. This type of tumor, left alone, may never harm a woman's health. But because the tumor is found in a mammogram, the woman undergoes surgery, and maybe also chemotherapy and radiation—all aggressive treatments with possibly serious side effects.

Olsen and Gotzsche think the studies show that mammography's disadvantages are greater than its benefits. They say that mammograms might save some women's lives by detecting cancer early. But they also think that mammograms actually end more lives than they save.

This is because many women have tumors that are very slow-growing and would not have any effect on their life if they weren't detected. But when mammograms detect these slow-growing tumors, the women receive aggressive treatments that are actually more dangerous than the tumors themselves.

Are Olsen and Gotzsche right? Did any of the studies they reviewed show an increase in overall deaths that could be linked to the consequences of having mammograms? This is where things really get messy.

Olsen and Gotzsche argue that overall death rates might actually be increased as a result of unnecessary stress and aggressive treatments (operations, toxic therapies) followingmammograms. The Cochrane editors argue that in order to find any kind of statistically significant change in overall death rates, a study would have to include more than two million women.

About half a million women were in the eight studies analyzed by Olsen and Gotzsche. But thei meta-analysis included far fewer women, because it combined only two of the eight studies. So, according to the Cochrane editors, there's no solid proof that mammograms are harmful, because there weren't enough women even in all of the mammogram studies combined.

What are the limitations of Mammography?

Although mammogram technology can be a very useful tool, it does have its limitations.

• First of all, only the breast tissue that protrudes enough to get onto the x-ray film will be imaged. This leaves unimaged areas of breast tissue that extend under the arm or perhaps the area deep against the chest wall. Additionally, benign tumors, malignant tumors, and areas of dense tissue appear white on a mammogram. This means that the presence of a tumor may go unnoticed in a woman with dense breast tissue.
• As stated previously, however, a woman’s breasts become less dense as she ages, and this less dense fatty tissue appears gray on mammography. Tumors are more easily visualized in this type of breast.

Despite the limitations of mammography, the probability of missing a breast lesion is only 8-10%. This is the reason why any screening type of examination should not only include mammography, but also breast physical exam, to detect the 10% of breast cancers that do not appear on mammography.

Some women may worry about the radiation absorbed during mammography. This is roughly the same amount of radiation to which a person flying over Denver would be exposed. The radiation exposure due to mammography is generally accepted to be negligible and should not be a deterrent for women to obtain yearly mammography.

Act 4: Same Studies, Different Researchers, Opposite Claims

Article published in The Lancet, February 2, 2002

To add to the confusion, a February 2002 article in The Lancet says that the very study Olsen and Gotzsche considered the most reliable DOES in fact show a large decrease in breast cancer deaths thanks to mammograms.

Are there any new data to support this different conclusion? No— just a different analysis of the same numbers. The authors of this article looked at one period of time within the study, during which there was a big decrease in breast cancer deaths for women over 55. They claim that this is the only segment of time that should be considered, because it's when the effects of mammography can actually be detected.

Is this claim rock solid? No. The benefits these researchers claim to have found might be due to something called lead-time bias.

What Is Lead-Time Bias?

There's one big problem with clinical trials on screening methods, in general. If you screen a large group of healthy people for a certain disease, you're likely to find people who have early stages of that disease. These people probably would not have been diagnosed until much later (when symptoms developed) if they hadn't been screened.

By diagnosing people earlier, you automatically add some years between the early diagnosis and later development of the disease, making it seem as if these people have lived longer than others who weren't diagnosed as early.

For example, let's say that without a mammogram a woman's cancer is discovered at age 60, and she lives well but eventually dies of the disease at age 75. She lived for 15 years following diagnosis. If she were screened, the cancer might have been discovered earlier, when she was 55. In that case, she would have lived 20 years following diagnosis—five years longer than without the mammogram. But she still dies at age 75.

So, lead-time bias is one explanation for why survival appears to be extended when mammography finds breast cancer early. The extra years added to the survival rate are actually only the years between when the cancer was detected by the mammogram and when it would have been detected without it.

Conclusions (For Now)

1. Clinical studies are open to interpretation.
   It's all in how you look at the numbers. So many factors affect how people respond to disease and treatment that it's very hard to design a perfect randomized clinical study. What does this mean about mammograms? That there probably will never be a completely reliable, indisputable study showing that mammography is beneficial or that it's harmful.
   
   So what should public health officials do about making recommendations regarding mammograms? Policy-makers have a tough job in this case. They have to review all the available evidence and the different interpretations of that evidence. Then they have to decide whether or not to change present guidelines. This is something that could take a long time.
    
2. Medical guidelines can change, but only if new evidence is found.
   Guidelines are just guidelines. They are somewhat arbitrary to begin with, and they can change over time with new information. Right now, there is no new evidence available on mammography, just new interpretations of old evidence. That's why it's so complicated, and that's why nobody's rushing to change any guidelines. In the future, mammography guidelines may change if solid new information becomes available.

In the meantime, remember that:

• The current debate is based on old studies. Any new look at old information is of limited value, because of the large number of possible interpretations.
• The field of mammography is changing and developing. Mammogram technology today is much better than it was 20 years ago. And mammography tomorrow will be even better.
• Mammograms are rarely used alone nowadays. If something potentially serious is found, other tests are used to find out more and to help determine the best possible treatment. This lowers the chances of false positives and unnecessary diagnostic procedures and treatment.
• It's unrealistic to think that any new study could answer the question of whether or not mammograms are beneficial. This is because now that mammograms are recommended for all women over a certain age, it would be unethical to ask some women to participate in the "control" group, which would not receive the possible benefit of screening. [top]

Take-Home Message

Breast cancer is the most common cancer to affect women, and mammography is the only screening method that's been extensively studied and shown to be able to detect breast cancer early.

The recent debate about mammograms is based on a new review of old studies. The review concludes that we can't say for sure that mammography saves lives and that we also have to recognize that mammography costs money, and can lead to extra biopsies and over-treatment.

Here's our take on this whole controversy: It's clear that mammography is far from perfect. Yes, today's mammography is better than ever before. But we still need a far better screening test that can do a much better job at distinguishing cancer from normal tissue in the breast, so fewer women experience the trauma of a "false positive" (a diagnosis of breast cancer when in fact there is no cancer).

For the women who do have breast cancer, we hope that a new and improved screening test will become available as soon as possible, so that the cancer can be detected in its earliest form when it has the highest chance of cure with minimal disfigurement.

Until breast cancer screening improves substantially, critics shouldn't trash the only method of screening we have. That's irresponsible. More time and effort should be devoted to discovering a better test.

Women understand that they must do the best they can with the best that's available. There are no guarantees in life. When your doctor orders a mammogram, it's with the hope—not a firm promise—that mammography may help save or extend your life.

What these researchers forgot to do is ask the women themselves how they perceive the relative benefits and side effects of mammography. Doesn't that matter too? We believe that most women would be willing to accept the potential harm of extra medical intervention in exchange for the possibility—however small—of having their life extended or saved by early detection.

Let's all push for a critical advance in the early detection of breast cancer. In the meantime, follow existing screening recommendations: annual mammography after age 40. Based on your own particular situation, your doctor may suggest that you start screening earlier.

Keep in mind that mammography is just a tool; it doesn't act alone. The ability of mammography to give you an edge on life depends on many different factors.

Here's how to make mammography work best for you:

• Get your mammogram done in the best place possible.
• Take advantage of other breast imaging studies as needed to further check a possible breast abnormality seen on the mammogram.
• Choose a doctor who has experience with breast health, to minimize the number of necessary biopsies.
• Take advantage of the latest core biopsy techniques that require only tiny incisions and are therefore less disfiguring.
• Visit breastcancer.org regularly to learn about breaking news on promising new imaging studies.

Thank you for your careful attention to this important issue and thank you for trusting us to give you the best information possible. There is nothing more important than your life. Knowing that inspires us, fuels our mission, and keeps us focused. [top]

A Final Friendly Warning

Beware the media hype on health issues!

If you hear something on the news that goes against everything your doctors have been telling you, call your doctor and ask about it. Or consult a reliable source of up-to-date medical information, like breastcancer.org. Remember: The daily newspaper or local TV news is generally not the place to go for even-handed reporting on health and medical issues.

Emily Conant, M.D. and Marisa Weiss, M.D.